Analogy in the Semitic Languages 9781463221614

Citation preview

Analogy in the Semitic Languages

A n a l e c t a Gorgiana

301 Series Editor George Kiraz

Analecta Gorgiana is a collection of long essays and

short

monographs which are consistently cited by modern scholars but previously difficult to find because of their original appearance in obscure publications. Carefully selected by a team of scholars based on their relevance to modern scholarship, these essays can now be fully utili2ed by scholars and proudly owned by libraries.

Analogy in the Semitic Languages

Abel Huizinga

l gorgias press 2009

Gorgias Press LLC, 180 Centennial Ave., Piscataway, NJ, 08854, USA www.gorgiaspress.com Copyright © 2009 by Gorgias Press LLC Originally published in All rights reserved under International and Pan-American Copyright Conventions. No part of this publication may be reproduced, stored in a retrieval system or transmitted in any form or by any means, electronic, mechanical, photocopying, recording, scanning or otherwise without the prior written permission of Gorgias Press LLC. 2009

1

ISBN 978-1-60724-563-6

ISSN 1935-6854

Extract from The American Journal of Philology, vols. 11 & 12 (1890;1891).

Printed in the LTnited States of America

III.—ANALOGY IN T H E S E M I T I C

LANGUAGES.

I.

T h e received opinion on the nature of analogy as a linguistic phenomenon, and on the relation of analogy and phonetic law may be stated as follows 1. T h e phonetic laws are absolutely without exception. There are not two classes of sound-changes, regular and irregular, systematic and sporadic. 2 In speaking of phonetic law, however, it must be remembered that the idea of law is conditioned by the sphere in which it works and the material to which it is applied. W e cannot speak of phonetic law in the same sense in which we speak of a natural law in physics or in chemistry. The student of linguistic phenomena should always take into account the individuality of the language-user. 2. Whatever cannot be explained by regular processes of phonetic law must, in the main, be due to the influence of analogy. Most, if not all, apparently irregular and exceptional forms which cannot be brought under any known phonetic law, or which seem to violate such laws, have been formed directly after the model of other forms without etymological consciousness, simply by the power of association. These two forces, viz. phonetic variation and formation by analogy, are regarded as the most potent in bringing about individual instances of linguistic changes. Thus Sievers, in his article on Philology in the ninth edition of the Encyclopaedia Britannica, speaking of these two forces, s a y s : " T h e y generally work in turns and often in opposition to each other, the former frequently tending to the differentiation of earlier unities, and the latter to the abolition of earlier differences, especially to the restoration of conformity disturbed by phonetic change. Phonetic change affects exclusively the pronunciation of a language by substitut1 See Misteli, Lautgesetz und Analogie, in Lazarus' und Steinthal's schrift für Völkerpsychologie und Sprachwissenschaft, X I 365-475. 2 Cf., however, B A L 98 2, A. J . P., V 1 7 1 .

Zeit-

472

AMERICAN

JOURNAL

OF

PHILOLOGY.

ing one sound or sound-group for another. Analogical change is confined to the formation and inflection of single words or groups of words, and often has the appearance of being arbitrary and irregular. It is beyond our power to ascertain whence analogical changes may start, and to what extent they may be carried through when once begun. All we can do is carefully to classify the single cases that come under our observation, and in this way to investigate where such changes are especially apt to take place, and what is their general direction." Starting with these general premises, it is the purpose of this article to study the operation of analogy in the Semitic languages, and to present, in a systematic way, the results of this study. In a study of this kind we are confronted by three questions : 1. What is the relation of analogy to the characteristics and structure of the languages in which it occurs ? Are its nature, its manifestations, and the scope of its application at all modified or conditioned by the well-known peculiarities of these languages ? 2. What individual instances of analogical formations are found in these languages ? 3. How are they to be classified ? In conducting our investigations we may start from the wellknown fact that the whole structure of language, in its grammatical forms and categories, is, in a sense, analogical. It exhibits the operation of what we may call constructive analogy. T h e working of analogy as a constructive force in building up groups of similar words and forms may be stated as follows : It is obvious that different stems, forming different words and presenting different sounds and combinations of sounds, are used to express different ideas ; and further, that different modifications of the same stem, whether by internal vowel changes, or by the addition of prefixes, infixes and suffixes, express the same idea under different modifications and relations. But that different words should undergo the same changes and receive the same additions in the form of prefixes or suffixes in order to express the same relation or modification of the original idea, is clearly the result of analogy whereby words are grouped together in classes, and within these groups the change which is applied to one is applied to all. This formation of groups or classes of words and inflectional forms, and the application of the same inflectional change (using this term in its widest sense) to every word belonging to the same group, are the result of analogy. Each group is governed by a prevailing

ANALOGY

IN

THE SEMITIC

LANGUAGES.

473

analogy, and each individual of the group is treated in its developments and its modifications to express different relations, in accordance with this prevailing analogy. T h e Semitic languages are peculiar in exhibiting with great clearness and fullness the effects of this constructive analogy by the regularity and uniformity of their structure. So, for example, in the inflection of the verb we find that the general analogy which is normally exhibited in the stems with strong and firm consonants holds good for all verbs, and the deviations from this model of the strong or regular verb are only modifications owing to the peculiar nature and feebleness of certain consonants. From the simple form of the primitives, called the Qal or first form, are formed according to an unvarying analogy in all verbs the verbal derivatives, sometimes called forms, or stems, or conjugations, each distinguished by a specific change or added element, with a corresponding definite change in its signification, such as intensive, causative, etc. In other languages where such formations exist they are usually regarded as new derivative verbs. But in the Semitic languages they are incomparably more regular than in the Indo-European languages. 1 In these cases we have no reason to suppose that the present uniformity had to contend with original diversity. It may have been so, but the presumption is that it was not so. But the case is different when we consider another marked uniformity in the structure of these languages, viz. the fact that all inflectional stems have, or are asssumed to have, three stem-consonants. A s the languages have c o m e d o w n to us, we find a striking uniformity of appearance, but we have reason to suspect that it is at the expense of original divergency. In this case we have an instance of analogy partly as a disturbing and partly as a constructive influence. There are indications that the number of tri-consonantal stems was originally much smaller than at present, but in the course of linguistic development bi-consonantal stems were made tri-consonantal by the addition of another consonant until finally the latter formed the majority. 2 A n d although we have reason to suppose that the inflection of bi-consonantal stems was originally to some extent peculiar and different from the inflection 1 See K a u t z s c h - G e s e n i u s , H e b r e w Grammar, 25th ed., L e i p z i g , 1889, § § 3 9 , 2 and 41. 2 Cf. L a g a r d e , Symmicta, I 122 (Gottingen, 1877); Deutsche Schriften (1886), 285 ; B i l d u n g der N o m i n a (1889), 215.

474

AMERICAN

JOURNAL

OF

PHILOLOGY.

of tri-consonantal stems, yet the preponderance of tri-consonantal stems was so strong that even those bi-consonantal stems which remained for the most part gave up their individuality, in various ways assuming the appearance of tri-consonantal stems and conforming to their method of inflection. 1 It may be assumed, then, as an established fact that the present uniformity in the appearance of the languages, namely, the predominance of tri-consonantal stems, is at the expense of original diversity. Still, even here, after the uniformity had once been established, analogy works as a constructive force in the further inflection of these stems. But leaving for the present the consideration of analogy and the analogical structure of the Semitic languages in this sense of the term, let us examine the subject of analogy in its narrower, more specific sense, in the sense in which the word is generally used by recent writers, such as Misteli and Sievers; let us see what instances of such analogical formations we have in these languages, how they may be most conveniently classified, and how they are related to the fundamental structure and characteristics of these languages. Whatever theory we may adopt as to the original form and constitution of the (so-called) weak verbs, this much at the least is certain, that in their present form they present the appearance of verbs regularly inflected after the model of the strong or perfect verb, modified, however, by the peculiarities of the weak consonants found in the stem. A d d to this the fact that in some of the Semitic languages certain consonants (e. g. in Hebrew the gutturals) have certain peculiarities which give rise to corresponding peculiarities of inflection of the stems containing such consonants, and all the apparent irregularities of Semitic verb-inflection are accounted for. These different peculiarities give rise to different classes of verb-inflection, according to the ordinary denomination, verbs '"13. »"¡i, n"S, etc. But knowing something of the nature of these weak consonants, something of the nature of the differences which distinguish these different classes of stems in their various formations and inflections on the one hand, and something of the nature of analogy as it is commonly understood, and as it is exhibited in 1 Compare Whitney, Language and the Study of Language, p. 302 sq.; Stade, Lehrbuch der Hebraischen Grammatik, §l2a, I and §142-144; Kautzsch, Grammatik des Biblisch-Aramaischen, §§45 and 46 ; Noldelte, Syrische Grammatik, §§41 and 57.

ANALOGY

IN

THE SEMITIC

LANGUAGES.

475

other families of languages on the other hand, we are led to expect the occurrence of analogy just here. W e are led to expect that the peculiarities which distinguish one class of weak verbs from another, the barriers, as we may call them, which separate the different classes from each other, should be entirely or partially disregarded and the different forms confused. A n d such we find to be the case. W e have not only the general fact that all the inflection of those weak verbs which were originally bi-consonantal in their stems is analogical, i. e. based on the analogy of the stems with three stem-consonants, but we have also a great many specific cases of analogy. W e find many individual instances of verbs of one class treated as if belonging to another class, and hence inflected after the analogy of that class, or even disregarding the weakness or peculiarity entirely and inflected directly after the analogy of the strong verb. W e find also a few instances where the strong verb is inflected after the analogy of the weak. A l l such cases of analogical formation, therefore, which affect the real or assumed stem of any word comprise one class with three subdivisions. C L A S S I.

Analogical formations in the inflection of the verb or in the formation of verbal derivatives with reference to the different classes of stems. Under this head we have three subdivisions: A . Confusion of the different classes of weak stems. B. W e a k stems after the analogy of the strong. C . Strong stems after the analogy of the weak. This law of tri-consonantality, if we may so call it, in the stems of words, is one of the most prominent characteristics of the Semitic languages, and this class of analogical formations which has just been considered is closely connected with this same characteristic, in that stems which in their original form did not have three stem-consonants are conformed to the analogy of regular, original, tri-consonantal stems. Another peculiarity of the Semitic languages is the function of the vowel and the use made of differences and changes in vocalization to differentiate various classes of words and to construct different inflectional forms. Thus, in Arabic, qatala is the type of the active transitive verb, qatila of the intransitive, and qutila of the passive. 1 ' C f . Lagarde, Bildung der Nomina, p. 7 ( Z D M G X L I V 536).

476

AMERICAN

JOURNAL

OF

PHILOLOGY.

We have also many phonetic processes whereby vowels are changed, lengthened, shortened, etc., in the various processes of inflection. These changes and variations are so light and delicate that we may expect some confusion at times, and such we find to be the case. Sometimes these phonetic processes are firmly maintained, enabling us to ascertain the law which governs them. But in a great many cases forms are transferred from one class to another, and changes take place which are not warranted by any phonetic law. All these analogical formations connected with the vocalization of the different words and forms can be comprised in one class. This gives us CLASS I I .

Analogical formations involving changes and confusion in vocalization. Still another peculiarity of the Semitic languages is their method of inflection and of the formation of derivative words by preformatives, informatives, and afformatives, and the resulting distinction between servile and stem-consonant. These formations and distinctions are peculiarly subject to confusion, and hence give rise to many analogical formations which may be comprised in one class with four divisions. CLASS I I I .

Analogical changes in the formative elements of words. A. Mistaking servile or formative element for part of the stem. B. Mistaking a stem-consonant for a servile. C. Analogical changes in the formative elements themselves ; influence of one formative element upon another. D. Addition of serviles and formative elements where they do not belong. I have thus shown, in a preliminary and provisional way, the possibility of the existence of three different classes of analogical formations affecting respectively the stem-consonants, the vowels, and the formative elements of the different words and forms. But when we consider that every Semitic word can be analyzed into these three elements, viz. its consonantal stem, its vowels, and its formatives, and that these three elements are in the main so strongly marked and so clearly separated, we see at once that this analysis has furnished us a basis for the classification of analogical formations which, though perhaps not so profoundly philosophical as some other modes of classification which might be

ANALOGY

IN

TIIE SEMITIC

LANGUAGES.

477

adopted, still has the merit that it is simple, practically convenient, and exhaustive, and most of all, that it presents these analogical phenomena in closest connection with the structure and the characteristics of these languages. The results of our study up to this point may be summed up in the following propositions : 1. The whole structure of the Semitic languages and all the formations and inflections of words are analogical, using the term analogy in its widest sense. 2. Using the term analogy in its restricted and more usual sense when speaking of it as a linguistic phenomenon, those cases of analogical formation which do occur are connected most closely with the structure and various characteristics of these languages, and thus they are easily accounted for, in fact they occur just where this structure and these characteristics lead us to expect them. W e find, then, these two factors in the structure of the Semitic languages: on the one hand a notable degree of regularity in the recurrence of certain fixed types of forms and in the application of inflectional modifications; on the other hand, certain deviations from these types and normal processes under the influence of a disturbing analogy. But when we look more closely into the nature and relations of these two factors it becomes evident at once that they sustain a most intimate relation to each other. They are not the result of separate and distinct linguistic forces, or to go back of the language to the mind of the language-user, of separate and distinct faculties of the mind. They are the result of the same law working under different conditions. A s a linguistic law we call it the law of the group. A s the product of the human mind we ascribe it to the power of association. B y the constitution of the human mind each word is felt to be a member of a distinct class or group, and not simply an isolated individual; and the whole philosophy of analogy as a linguistic phenomenon may be thus briefly expressed : Whenever the law or type of any group has been able to maintain itself, the individual conforms to the law of the group to which it belongs and the formations are normal. But when the law breaks down and the distinction between the groups is disregarded, the individual, instead of conforming to the group to which it belongs, is transferred to another group, with which somehow it has become associated, and the result is a disturbing analogy.

478

AMERICAN

JOURNAL

OF

PHILOLOGY.

It is true that the unwarranted application or extension of analogy beyond its legitimate bounds is a marked feature of language. It is this that has given rise to the expression mistaken or false analogy. It might better be called disturbing analogy. One of its most frequent effects, as is pointed out in the words of Sievers, quoted above, is the obliteration of existing differences. But it is wrong to regard analogy simply as a disturbing influence. In fact, we might almost say that this is only an incidental effect. It is better to regard analogy as that which binds together the different individual members of each group of words or linguistic elements. Or to speak more accurately, it is natural for the mind to associate words which although different still are felt to be members of one and the same group, and to apply to all the inflectional changes which it has been taught to apply to one. This extension of the same inflectional changes to all the different members of the same group, this formation of different words from different stems after the same type or model, is not the work of memory, as is also pointed out by Brugmann in his book " Z u m heutigen Stand der Sprachwissenschaft," p. 79. It is the work of the power of association or combination, or, as it might also be called, the analogical faculty. W e are not concerned primarily with the name of this faculty, however. But that with which we are concerned is the fact that so-called analogical formations are not to be regarded as something isolated and entirely distinct from those formations which are called normal, but rather they are the results of the same mental process, and show the operation of the same law working under different conditions. Starting from the principle that analogy is the modification of an existing form, or the origination of a new form after the model of another form with which it is associated, different schemes for the classification of such analogical phenomena have been proposed. These different schemes are presented and reviewed b y Delbrück, " Einleitung in das Sprachstudium," p. 108 fg. According to him these classifications are made from three different points of view: 1. According to the nature of the psychological activities which are concerned in the different formations. 2. According to the nature of the words affected by analogy. 3. According to the results of the operation of this force of analogy. The first of these principles is the one adopted by Misteli in

ANALOGY

IN

THE SEMITIC

LANGUAGES.

479

the article already referred to (Zeitschrift für Völkerpsychologie und Sprachwissenschaft, XI 365-475, X I I 1-26), and after him by Wheeler, in his treatise on "Analogy and the scope of its application in language." 1 As this is by far the most satisfactory of existing methods of classification, I shall present it in some detail, using mainly the phraseology of Wheeler in the statement of the different classes, and illustrating them by examples taken from the domain of Semitic. According to Sievers, the influence of analogy tends largely to the abolition of earlier differences. Similarly Wheeler (op. cit. P- 35) : " The operation of analogy in language is in every case ultimately conditioned and determined by the natural quest of the mind for unity to replace multiplicity, system to replace anomalous diversity, and groups to replace monads. The office of the psychical factor in the development of language is therefore the maintenance and re-establishment of the groups which the phonetic laws tend to disrupt, and the creation of new groups. It aims to eliminate purposeless variety." To this must be added the important statement on page 37 of the same work : " The formation of new groups very commonly appears as a readjustment of old groups. Changes in the character and use of the phonetic material of a language often cause a form or number of forms to be severed from one group and attached to another." Wheeler thus classifies the phenomena of analogy: I. Likeness of Signification and Diversity of Form. Two words entirely diverse in form, but which are capable of application to one and the same object or idea, may, through the influence of this limited likeness of signification, be confused into one word by the process known as " contamination of form." No instance of this kind in Semitic has occurred to me. II. Affinity of Signification and, Diversity of Form.. Words totally dissimilar in form, but expressing ideas of like category, are made to approximate slightly in form through the extended application of some sign of category or through the extended use of some element or combination of elements of sound which has come to be recognized as characteristic of a group. As an illustration of this kind of analogy the following has occurred to me : The common people often say masoner for mason. The ending er came to be regarded as expressing trade or calling, 1

Ithaca, N. Y „ 1887.

Cf. also A . J . P. V 165-85, X 202.

480

AMERICAN

JOURNAL

OF

PHILOLOGY.

from such words as painter, joiner, carpenter, farmer, etc. Hence they sought to bring mason into the same category of form b y adding er and making masoner. Instances in Semitic are the following among others: T h e prefix m, to form the passive participle of the Arabic first form, of which the ground-form is qatAl. See Barth, Nominalbildung, p. 178, 1123a. 1 In Arabic this principle often works in the extension of existing groups. See Barth, Nominalbildung, p. 135, §85^. T h e form qaiil, formed nominally from the i- imperfect, is used to denote masses, collectives. Then in that sense it is formed from stems which have no i- imperfect, especially to denote masses or collections of animals, kalib ' dogs,' ' sheep,' etc. Cf. also the Mandean pronouns achnun, achtun, achtdchun and achn&chun (Nold., Mand. Gramm., p. 86). III. Likeness of Function and Diversity of Form. Words differing in form are reduced to groups upon the basis of likeness of function, i. e. of likeness of use in the economy of the sentence and for the expression of like modifications of thought. A n illustration of this kind of analogy is found in the formation of the Hebrew infinitive absolute of the Niphal. See Barth, op. cit. p. 72, §49^. The regular infinitive absolute of the Niphal is seen in "pll- But as in the Piel and Hiphil a certain assonance was perceived between the imperfect and the infinitive absolute used to strengthen it, e. g. -las^n Ex. 23, 24, a similar form was originated in order to produce a similar assonance in the case of the Niphal rn3R roan, n.^ps absn, I Sam. 27, 1, etc. This analogical formation almost entirely displaced the organic. Other illustrations of this are seen in the formation of prepositions with suffixes in Hebrew and in Ethiopic. Compare T h e 1 is organic in the first two cases, because they are to be referred to the stems 'Stf and Hi?. Starting from such stems it is extended analogically to stems of other prepositions where it has no place at all. IV. Contrast of Signification and Partial Likeness of Form. W o r d s of contrasted signification and of partly similar form are grouped in couplets, and a further approximation in the outward form is the result. 1 Cf., h o w e v e r , Beitr. z. Assyriologie I 160 ( Z A , I V 375). 2 C f . Z D M G , X L I I part 3 (A. J . P . X 234) and L a g a r d e , M i t t h e i l . I I 231.

ANALOGY

IN

THE SEMITIC

LANGUAGES.

481

A n instance of this kind of analogy is given b y Praetorius, Ethiopic Grammar, p. 86, §99. Eth. watf'a, he has gone out, has in the subjunctive idd', and imperative (la'. T h e s e forms ought to be inflected thus: ted'e'i, ide'A or del, de'd. T h e y follow, however, the analogy of their opposite ib&\ from b6'a, he has gone in, and hence we have the forms tcdai, idafi., idd'd, etc. 1 A n o t h e r illustration of the a n a l o g y of the opposites is the H e b r e w jfPj?, last (stem f S p ) after the analogy of ¡ispay first. V . Likeness of Signification and Partial Likeness of Form. W o r d s whose stems have a like signification and are similar but not like in form are grouped together upon the basis both of meaning and form, and a levelling of the form of the stems is the result. Under this head W h e e l e r brings the following classes of analogical formations: A . L e v e l l i n g between different cases of like stems. A s an instance, somewhat similar at least to the instances mentioned b y W h e e l e r under this head, may be mentioned such cases as I have given under CI. I l l , Div. B., the Assyrian plural ii&ti for iddti, from ittu, which stands for idtu,2 the feminine of idu, hand. B. Levelling between the different forms for person and number in the same tense of the verb. Analogical influence of this kind is v e r y conspicuous in the inflection of the perfect of the Semitic verb. T h e original paradigm probably was as follows, in the singular :3 qatal a. ' he has killed.' qatal at. ' she has killed.' qatal ta. ' thou hast killed.' qatal ti. ' thou (fem.) hast killed.' qatal kti. ' I h a v e killed.' In Ethiopic, the first person qatal-kil has influenced the second person masculine and feminine, so that they now have qatalkd, qatalki. In the other languages the reverse of this process took place ; kH of the first person was changed to id under the influence of the a n a l o g y of the second person, while in H e b r e w , b y a further analogical change, tii was changed to it under the influence of the possessive suffix i.1 Cf. Hebraica I I 6, i. See Hebraica I 178, 5 ; cf., however, Delitzsch, Prolegomena 46 and 115. 3 See Noldeke, Die Endungen des Perfects, Z D M G , X X X V I I I 407 sq. 4 C f . S F G 53 below ; Z D M G , X X X V I I I 419; X L I V 539, 1 ; Wright, Comp. Gramm. 175. 1

5

482

AMERICAN

JOURNAL

OF

PHILOLOGY.

C. Levelling between the different parts (tenses), etc., of the same verb. An instance of this kind of analogical formation is seen in the vocalization of the Hebrew perfect and imperfect Piel as compared with the corresponding forms in Arabic. In Arabic, which probably comes nearest to the primitive, we have qaiiala, yuqattilu, while in Hebrew we have 'jap, The ¿»-vowel of the second syllable of the perfect is probably due to the influence ol the corresponding vowel of the imperfect. The /"-vowel of the first syllable is then perhaps due to vowel-harmony. D. Levelling between derivative and primitive. No instances of this kind in Semitic have occurred to me, perhaps because there generally exists such a close connection between derivative and primitive. For this classification Wheeler claims practical exhaustiveness. He says : " Under the five main categories which have been thus far established and discussed may be classified nearly if not quite all the phenomena usually associated with the action of analogy." I have preferred, however, in my treatment of analogy to look at the phenomena from a different point of view, and adopt a different method of classification. The predominant characteristic of all Semitic analogical formations, as I have already shown, is the formation of groups, and the disturbance of these groups by transferring individual words, forms, portions, or characteristic elements of forms from one group to another, taking them from a group where they exist organically and applying them to or placing them in a group where they do not belong. These disturbances or transferments will naturally affect either the consonantal ground-stem of a word, or the vocalization of the stem, or those formative elements, prefixes, infixes, suffixes, by which different inflectional forms or derivative words are formed. Hence the classification which I have given will logically follow, and I trust that a study of the material as I have arranged it, in the second part, under these different heads, will bear out its suitability. This material will appear in a subsequent number of this Journal. ABEL H . HUIZINGA.

I I . — A N A L O G Y IN T H E S E M I T I C L A N G U A G E S . PART II.

The following are the principal instances of analogical formation in the Semitic languages. C L A S S I.

Analogical formations in the inflection of the verb, or in the formation of verbal derivatives with reference to the different classes of verb-stems. In this class are comprised only such cases as have reference to the different classes of verb-stems. All other cases are placed under their proper heads (Classes I I and III). Division

A.

Confusion of the different classes of stems, either in the inflection of the verb or in the formation of verbal derivatives. In Hebrew the two classes of verbs n'6 and X'6 are frequently confounded. Sometimes, however, this confusion is due simply to the Massoretic pointing. n"1? after the analogy of X'6. 2 Kings 25. 29 ' he changed ' for IW. So also from the same stem Lam. 4. 1 and Ecc. 8. 1. an?! 'he is fruitful' Hos. 13 15! 'to consume' Dan. 9. 24 for (inf. constr. Piel). KSnp Jer. 38. 4 ' relaxing ' for nSlD (prtcp. Piel). 'IK"!» for mb Zeph. 3. 1 'rebellious' (prtcp. Qal) st. m». D'X^Dp Lam. 4. 2 ' weighed ' as if from Deut. 28. 66 ' hanging ' pass, prtcpl. Qal as if from }On. So also 2 Sam. 21. 12 np. K^D?.! 2 Chron. 16. 12 'and he was sick' for r6m. IXSn^l 2 Kings 17. 9 'and they covered' as if from astern KDn, for nan. 2 Kings 17. 21 3*riD 'and he removed' as if from SHJ, for ill from m i .

ANALOGY

IN

THE

SEMITIC

LANGUAGES.

31

K^FT^K Prov. 1. 10 ' b e not willing' as if from a stem SOX for m x . 2 Chron. 26. 15 ' t o shoot' as if from a stem NT» for m \ similarly 2 Sam. 11. 24 and Prov. 11, 25. X"? verbs after the analogy of verbs. Several forms from 'to withhold,' ' t o restrain,' are formed as if from a stem r t a , so ' n t f ^ Ps. 119. 101 (this, however, may be simply due to the Massorites), ^n^? 1 Sam. 25. 33, I Sam. 6. 10, n ^ j Gen. 23. 6. So also many forms from ' t o heal' are formed as if from ¡IS"}, as ns"! Ps. 60. 4. Imperative Qal with vowels as if from xa-i. WS-jF) Job 5. 18 'they heal.' n n | l 3 Jer. 51. 9. 2 Kgs. 2. 22. Perfect and imperfect Niphal. -ISTJer. 8. 11. Piel. So also npi3 Prov. 12. 18 'babbler' from stem KD3 (cf. X??*?). nPV R u t h 2. 9 from NP? ' t o thirst.' • t e Ezek. 28. 16 and n ^ J o b 8. 21 from «!?» 'to be full' as if the stem were n1?». Compare also such infinitive forms as ntK"H Jud. 8. 1 from Kip ' t o call.' riK^p Numb. 6. 5 from 'to be full,' also ms6p Jer. 25. 12, n ^ p E x . 31. 5. n w b Prov. 8. 13 from ' t o hate.' n s b q Ez. 33. 12 from ' t o sin,' n'K3p_ 2 Sam. 21. 2 from K.3i? ' t o be jealous.' These forms as compared with N^P, etc., clearly show the influence of the analogy of T h e example of J"iNt3q, moreover, leads us to suspect that in Ntairt Is. 65. 20, Ntin Ecc. 8. 12, and in other similar cases we have something a "confusion due to the Massorites." 1 W^aJ Ps.more 139. than 14 and -13 Ex. 33. 16 are from t6a to separate, distinguish,' as if the stem were r6a. W'DtJJ Job 18. 3 from NPP 'to be unclean' as if the stem were HDD.

So IJVVPn 2 Sam. 3. 8 from nm

' t o find ' as if the stem were

1 It is well known that the verbs K" 1 ? are in the Mishna regularly inflected as verbs see Geiger, Lehrb. zur Sprache der Mishna., p. 46; StrackSiegfried, Lehrb. der neuhebr. Sprache (1884), §§g8c and 1 0 5 ; cf. Wright, Koheleth (1883), p. 488.

32

AMERICAN

JOURNAL

OF

PHILOLOGY.

n|OJ Jer. 49. 10 from Kin ' t o hide' as if the stem were nan, also rnnn 1 Kgs. 22. 25. So also JV33 Jer. 26. 9 and WiOJn Zech. 13. 4 from N3J 'vaticinatus est* as if the stem were H33 (for 133). Similarly ni3JJ}n 1 Sam. 10. 13 and 1 Sam. 10. 6. Again, a similar confusion is observed between V"v stems and 1"!' stems. The following forms from V"V stems are formed as if from a stem ?"i?: I'll* Prov. 29. 6 ' h e will sing' (st. |3i). f n ; Is. 42. 4 ' h e shall be bruised' (stem ^Xl). [Cf. Eccl. 12. 6.] "ME* Ps. 91. 6 ' h e shall waste' (st. Tie»). >12 Ecc. 9. 1 ' to explore' (st. TO). •1T1J3 Nah. 1. 12 'they shall be cut down' (st. i?j). H3H Is. 24. 3 'she shall be plundered' (st. ??3). pBR Is. 24. 3 ' she shall be emptied' (st. pp3). Ps. 115. 7 'they handle' (st. WD). T 9 5 Jud. 3. 24 'covering' (st. ~pD). Hab. 3. 9 ' she was made bare' (st. Til?). IDTi; Hab. 2. 17 'it terrifies t h e m ' (st. nnn). Prov. 11. 15 ' h e shall be broken' (st. l?m). So also in Syriac these two classes of verb-stems, viz. ¡}"S and V'l? are frequently confused. See Noldeke's Syriac Grammar, §§58. 105. 126B, 174E, 175, 176, 177 A for these and similar instances in Syriac. The following forms from VI? stems are formed as if their stem were V"V. 3K> Josh. 2. 16, inf. of 3-1E> 'to return'; 103 Jer. 48. 11, Pause, ' h e is changed,' st. 11D; and "iPJ Ezek. 48. 14, Hiph. from the same stem. In Syriac all the verbs V"V form their forms with prefixes after the analogy of verbs J"S. See Noldeke, Syriac Grammar, §i78C. This is also the explanation of all those forms with prefixes from verbs !?"l? in Hebrew, in which the first stem-consonant is doubled. They are based on the analogy of verbs J"S. Instances are DM* from Dm 'to be silent.' "»P! stem Tip 'to bow the knee.' M? stem DDH 'to finish.' stem 'to be slack, feeble.' n^! stem nna> 'to be low.' T?' stem D 'to tumble, fall.'

ANALOGY

IN

THE

SEMITIC

LANGUAGES.

33

3D? stem 33D ' t o t u r n ' a n d 25!, Hiphil from the s a m e stem. T h e s a m e formation is o b s e r v e d in the case of s o m e forms from Y'V stems. SSI stem MD ' t o d r a w b a c k , depart.' stem SlO ' t o cut off.' W n Hiphil stem Sir ' t o s h a k e . ' rVE>! Hiphil stem n-1D ' t o m o v e . ' Va; H o p h . stem SiD ' t o b e l o n g . ' H'Sn a n d ' t o l a y d o w n , ' ' t o cause to r e s t ' H i p h . stem n-U. A c c o r d i n g to S t a d e , §535b, the form

impft. 3d person

fem. plur. stem 33D ' t o turn,' with the insertion of the d i p h t h o n g ai ('v) between the stem and the e n d i n g , rests on the a n a l o g y of the s a m e form from stems n'6 (for or v 6 ) i i j ^ 1 ? . In M a n d e a n ( N o l d e k e , M a n d e a n G r a m m a r , p a g e 83) a n d in A m h a r i c (Praetorius, A m h a r i s c h e S p r a c h e , p. 1 4 1 ) confusion b e t w e e n stems T'T and a n d Y'b is especially frequent. The e x a m p l e s a r e all g i v e n in the places cited, so that it is needless to g i v e them here. A c c o r d i n g to Praetorius (loc. cit.), this is also the e x p l a n a t i o n o f the form ni3D, i. e. the stem 33D is treated as if it w e r e H3D for 13D ( s a b a w a ) . O n nt3D s e e also W r i g h t , A r a b . G r a m m . §120 rem. c. a n d Spitta, D i a l e c t of C a i r o , §101, 3. S o m e t i m e s stems V'B a n d »"B are c o n f u s e d . A s is well k n o w n , m a n y of the stems '"3 in H e b r e w are o r i g i n a l l y V'B, the original initial wazv r e a p p e a r i n g in the N i p h a l , H i p h i l , and in verbal derivatives with » - p r e f o r m a t i v e . N o t all s u c h formations w i t h waw a r e organic, h o w e v e r . T h u s , as is pointed out b y H a u p t ( S . F . G . origip. 22, note 1) the v e r b ^ ' t o k n o w ' is a v e r b cum yddh nario, as a p p e a r s from the A s s y r i a n idi, tidi, and E t h i o p i c 'ayde'a nolum fecit. H e n c e the H i p h i l JHin, S y r i a c 'awda' and shawda' are analogical formations. S o m e forms from stems V'V s h o w t h e influence of analogy to w h i c h t h e y bear an e x t e r n a l r e s e m b l a n c e . S o -nX! J o b 18. 7, impft. stem m * ' to b e narrow.' T h e v e r b ^ n ' t o g o ' has the* impft. Q a l . and H i p h . T ? i n as while the form na^K Mic. 1. 8 points to a if the stem were form cum ' originario. S o also the form impft. from 310 ' t o b e g o o d , ' as if the stem w e r e 3D\ In A s s y r i a n the s t e m s V'B, K"B and n"B (K,B> are also c o n f u s e d . T h u s ustbila, stems N"B.

use sib, usisi

f r o m stems V'B are f o r m e d as if from

34

AMERICAN

JOURNAL

OF

PHILOLOGY.

So also attasab, ittasAni, muttabil are formed after the analogy of attalak, etc., and this again shows the influence of stems ¡"3. izsazd, present of na.za.zti' to stand,' is based in its vocalization on the analogy of 'illakli from 'al&ku " to go.' See Haupt, S. F. G., p. 52, note 10. So also iddan stem nad&nu 'to give,'by the same analogy. illika 'he came' is J"B analogy. See Haupt, S. F. G., p. 66, also Hebraica, Vol. I, p. 255. So also the stems "Ol and pn form their preterites as if from ip' and p"i\ e-qir, li-ri-qu (Del. Ass. Gram. §112). Dr. Rosenberg, Das Aramaische Verbum im Babylonischen Talmud, gives the following instances of transfer from one weak class to another in that idiom. P. 40. "1XJVD ' er ward gebildet' Joma 85a, Sota 45b analogy Of 1"!?. P. 44. Most verbs l"U form their Afel after the analogy of verbs IB. P. 45. Verbs V"V with object-suffixes are sometimes treated after the analogy of I"1?. P. 46. Ethpeel of verbs i?"l?, formed, partially at least, after the analogy of verbs VI?. So bhv forms its Pael and Ethpaal after the analogy of verbs VI?. P. 63. Verbs form with pronominal suffixes after the analogy of the strong verb (cf. Nold. Mand. Gram. §204). In modern Syriac NIVB, fem. xnnB 'lukewarm,' whose stem is really niB, is based on the analogy of KTp 'cold,' XDTl 'warm' (Nold. Neusyrische Sprache, §43). Further, KURDS' 'heard' and K'lTiD 'sick' are formed after the analogy of other »"*? forms, such as 'pure,' Kin 'seen,' etc. (Nold. Neus. S. §44, p. 91). Cases of analogical formation in the confusion of different classes of weak verbs are specially frequent in modern Syriac. See Nold. Modern Syriac Grammar, p. 188, §95. Verbs V'S and VI? are confused. See Nold. op. cit., p. 228, §108, p. 230. Verbs K"S, such as pDK 'to ascend,' 1DK 'to catch, to be cold,' are treated after the analogy of verbs V'B. Many verbs originally 1?"1? are treated after the analogy of verbs VI?, e. g. e|to, KB'D 'to bend, stoop' from fjBa, DKrt, etc. 'to finish' from DDH (Nold. Neusyr. Sprache, p. 231, §109). 2ND 'to be worth' has some forms from iOD O"1?).

ANALOGY

IN

THE

SEMITIC

LANGUAGES.

35

' to spit' has in the inf. NpliU after the analogy of N"2 (Nold. §110, p. 233). W'h and are confused (p. 239, §111). See also p. 248, §114. T h e verb an* (p. 254, §117) forms its conjugation from several different though correlated stems. T h e r e are many instances of such confusion of the different classes of weak verbs in Mandean. T h e instances are too numerous to be mentioned here in detail. See Nold. Mandean Grammar, p. 24, §22. V e r b s tertiae 1? after the analogy of verbs tertiae \ P. 82, §74. Verbs V'fl after the analogy of V"V and conversely after the analogy of v'l?. Verbs mediae K and 1? after the analogy of W. Verbs after the analogy of V"V, and so conversely verbs V"0 after the analogy of verbs y 'b. Nominal formations from different stems showing the influence o{y'h stems are given on p. 104, §90. P. 111, §94. (Mand.) forms from »'6, v 6 treated after the analogy of forms from stems Sl"V. P. 236, §i77a. V e r b s terliae gutturalis after the analogy of stems V'1?. P. 243, §180. Verbs are treated after the analogy of verbs VB, '"a. V e r b s 1"1> and V"V are habitually confounded (Mand. Gram, p. 247, §183). S o p. 255, §189. Verbs mediae 1) or it are treated after the analogy of verbs l"r. Cases of the influence of one class of weak stems upon another in Amharic are given in Praetorius, Amharische Sprache (Halle, 1879), p. 141, §103. A s they are all fully cited there and discussed at some length, it is needless to mention them here in detail. Other cases of the confusion of different weak stems in Arabic are given by Barth, Nominalbildung, p. 45, §3ob. T h e adjective-form qatil from stems 1"I>, preserves the w in a few cases such as tawil' long,' hawid ' zealous.' But in most cases they follow the analogy of stems with the usual change of &jt

to ajji

('aiii).

Examples

are maiiit

' d e a d , ' 'aiiid,

' g o o d , ' haiiin 'light, easy.' Other instances are given p. 188, §i27c, qaiiim, haiiit,

jaiiid

saiiid.

36

AMERICAN

JOURNAL

OF

PHILOLOGY.

C L A S S I.

Division

B.

Stems with weak stem-consonants after the analogy of stems with strong stem-consonants. Also original bi-consonantal stems after the analogy of tri-consonantal stems. As was pointed out in the first part of this article, many stems in the Semitic languages which seem to have three stem-consonants were probably bi-consonantal in their original condition, and assumed an additional stem-consonant in order to make them tri-consonantal by the force of the prevailing analogy. Other stems, instead of becoming completely tri-consonantal remained more or less imperfect and preserved some traces of the original bi-consonantal condition. In different languages also they attained to different stages of completeness. Thus, from the original biconsonantal stem Dp we have in Hebrew Dip and EpT (with a long vowel), and in Arabic q&ma. The Hebrew intensive is DDIp, and the Arabic qauiiama, and Aramaic Dip (Hpjp1? Dan. 6. 8). Here it will be seen that the original bi-consonantal stem is more conspicuous in Hebrew, and the form of the tri-consonantal stem is more completely attained in Arabic and in Aramaic, qauuama standing on exactly the same plane as qattala. Finally, taking the inflection as it now stands, where the peculiarity or feebleness of one or more of the stem-consonants, or the originally bi-consonantal nature of the stem gives rise to different classes of verbs, each with its own special paradigm (according to the current denomination '"a. ¡"2' ii"1?• etc.), we find many individual cases which, disregarding these peculiarities, leave the paradigm to which they ought normally to conform and follow the analogy of the strong or perfect verb throughout. W e find, then, that we can distinguish three closely related cases: (1). Stems in which all traces of the original bi-consonantal nature have disappeared, and which have three stem-consonants in all of their forms. (2). Stems in which the analogy of the tri-consonantal stems is established as part of the regular paradigm in some of the forms. Here we find such cases as the Arabic qauuama from q&ma, Aramaic from Dp. Compare also Noldeke, Modern Syriac Gramm. §42, where it

ANALOGY

IN

THE

SEMITIC

LANGUAGES.

37

is o b s e r v e d that t h o s e nouns w h i c h in the earlier l a n g u a g e w e r e bi-consonantal, s u c h as dim ' b l o o d ' a n d shim ' n a m e , ' h a v e b e c o m e tri-consonantal in m o d e r n S y r i a c , dimmd, a n d shimmd. U n d e r this h e a d m a y a l s o b e p l a c e d the f o r m s t r e a t e d b y B a r t h , V e r g l e i c h e n d e S t u d i e n , Z . D . M . G . , V o l . 4 1 , p. 6 0 3 f g . B e c a u s e of the r e l u c t a n c e of the S e m i t i c l a n g u a g e s to h a v e n o u n s with o n l y t w o s t e m - c o n s o n a n t s , m a s c u l i n e n o u n s of this class of f o r m s of w h i c h h e is treating (viz. s h o r t e n e d d e r i v a t i v e s of s t e m s and a r e but s e l d o m f o u n d in the northern S e m i t i c l a n g u a g e s , and in A r a b i c h a r d l y e v e r . I n c o n s e q u e n c e of the p r e v a l e n c e of the law r e q u i r i n g three stem-consonants, the g r e a t e r n u m b e r of s u c h s h o r t e n e d nouns a s s u m e d the feminine e n d i n g , not to d e n o t e s e x , but s i m p l y to c o m p e n s a t e the loss sustained b y the d i s a p p e a r a n c e of the third s t e m - c o n s o n a n t . Cf. H e b r e w ' d r i n k ' f r o m the stem npt5>, n}J ' b o d y . ' I n A r a b i c hize f r o m the stem haziia ' t o o b t a i n ' and m a n y others. S o also the " f e m i n i n e e n d i n g " is a d d e d in those cases w h e r e the first s t e m - c o n s o n a n t has d i s a p p e a r e d . and Arabic lidatun f r o m the s t e m "6l (walada),~h°l 'to bear. 'This " f e m i nine e n d i n g " has b e c o m e s o entirely a p a r t of the stem as to r e m a i n in the plural. C o m p a r e J u M h . ' d o o r s , ' HW^P ' b o w s , ' nihps? ' w a t e r i n g - t r o u g h s , ' n i n p ? E z . 1 3 . 1 8 ' b a n d s , pillows.' ( 3 ) . I n d i v i d u a l cases o f s i n g l e f o r m s w h e r e the n o r m a l p a r a d i g m or t y p e of the class to which the s t e m b e l o n g s is disreg a r d e d a n d the f o r m is b a s e d directly on the a n a l o g y of t h e s t r o n g or r e g u l a r v e r b . I n T 1 ?.!* E x . 23. 22 ' I w a s a n e n e m y ' a n d in 'weary,' we e v i d e n t l y h a v e a later d e v e l o p m e n t of a consonantal between the t w o original c o n s o n a n t s of the stem. 1 W i t h r e g a r d to h o w e v e r , w e m u s t note that B a r t h , N o m i n a l b i l d u n g , § i o b , r e g a r d s it as t r a n s p o s e d for T h e assimilation of a v o w e l l e s s t is a f a m i l i a r fact in H e b r e w m o r p h o l o g y . I n cases w h e r e it is r e t a i n e d , the a n a l o g y of other c o n s o n a n t s not subject to s u c h assimilation d o u b t l e s s has its influence. S o i ' ^ n I s . 58. 3 ' y e e x a c t , ' "ilBf J e r . 3. 5 ' H e will k e e p , " -11S3? D e u t . 3 3 . 9 ' T h e y will g u a r d , ' 3p}? J o b 40. 24 f r o m ' t o b o r e t h r o u g h , ' -13p?.11 Is. 29. 1 f r o m S]pJ ' t o c o m e r o u n d , ' said of feasts. 1 Cf. Haupt, Zeitschrift für Assyriologie I I 276, and Beiträge zur Assyriologie I 296 below.

38

AMERICAN

JOURNAL

OF

PHILOLOGY.

TJ?}']^ E z e k . 22. 20 ' t o melt,' Hiph. inf. of^DJ. pn?n Jud. 20. 31 H o p h a l from PO? ' t o cut off.' In Assyrian n is not assimilated to the following dentals: enzu 'goat,' bintu 'daughter,' ensu 'weak.' See Haupt in Hebraica I 227. In Ps. 1 1 9 . 6 1 ' t o surround' Piel of n-H'; "M ' t o make blind,' P s . 1 1 9 . 28 'preserve me alive !' PielofD-1p; Dan. 1 . 1 0 ' s o y e should endanger,' Piel of the stem 3-1H; Josh. 9 . 1 2 ' w e provided ourselves with food,' Hithpael denominative from T V ' food,' from the stem 1-1V ' to hunt,' we have instances of forms of stems with weak stem-consonants treated after the analogy of stems with strong consonants. In a few cases we find, instead of the usual form of the construct infinitive from n"1? stems (01^), a form which is based on the analogy of the same form from the strong stem

; such

forms are nbl? ' t o do,' ntq ' t o see,' njp ' t o get,' ins? ' t o drink.' Such forms as D n i f 'boves,' D'mq 'thickets,' O ^ q 'forces,' D 1 ^"! 'bucks,' 'cities,' when compared with the more organic form of such stems as seen in CpiC? 'whips,' O'VP 'thorns,' 'terebinth-trees,' show the influence of the analogy of the form nna^. In the different forms of the verba mediae geminatae (or V"V verbs) there are many forms which seem to point to an original bi-consonantal stem, e. g. 3D, while others are formed as if there were three original firm and equally legitimate stem-consonants, 33D. T h e former are the more organic, more in accordance with Hebrew phonetic l a w ; the latter are analogical. T h u s by the side of such forms as P I ' he has beaten small,' DO ' h e is warm,' and ID 'it is bitter,' we have other forms with the second stem-consonant repeated. T13 ' h e has plundered,' ' h e is warm,' "HI? ' h e has measured,' 33D ' h e has surrounded.' After the same analogy we have substantives such as 0?? ~ ' terror,' the plural D'pn? ' nations " by the side of the more organic form D^». S o also in the construct infinitive we have such forms as tT| ' t o cut,' 33P1? Numb. 2 1 . 4 ' t o compass,' TIT? ' to make strong,' HIV ' t o bind up,' DDJp Is. 10. 18 ' t o melt away,' "rnp ' t o s p o i l ' Jer. 47. 4, ^ i f 'to spoil,' by the side of the more organic forms 3b, tW. ri. S o also the imperfect TH' (stem T13) ' to flee f r o m ' and P.m f r 0 m the stem pn ' t o be gracious,' by the side of the more organic t'nJ. 1

Cf. Noldeke, Mand. Gramm. §132 ; Syr. Gramm. § § 2 l D and 93.

ANALOGY

IN

THE

SEMITIC

LANGUAGES.

39

In Hebrew all forms with the (so-called) Dagesh forte impiicitum, especially in the Piel, Pual, and Hithpael of verbs whose second stem-consonant is a guttural are analogical. The phonetic law is that the guttural cannot be doubled. Then the short vowel preceding the guttural must be lengthened because it stands in an open unaccented syllable. Cases where this is not done, e. g. ' he has consumed,' KB? ' he has despised,' • he has rejected' are based on the analogy of^i?. The "virtual doubling" of the gutturals is a fiction of the grammarians. CLASS

I.

Division C. Forms of the strong stem after the analogy of forms from weak stems. One of the most remarkable cases where the form of the strong verb has been influenced by the weak verb in Hebrew is the long accented /-vowel in the final syllable of the Hiphil or causative. So Bickell, §47 : " The remarkable transition of i, which has arisen from a to i in the Hiphil, has perhaps originated according to an erroneous analogy from the conjugation of the verbs mediae v.J., where this i is phonetically legitimate." Compare also Stade, Hebrew Grammar, §91. König's criticism (Lehrgebäude, §27. 4), that the basis is too narrow, and that it is too bold to explain a form of the regular verb by a single form of the irregular verb, is hardly borne out when we consider the prevalence and power of analogy. See Delbrück, Einleitung in das Sprachstudium, p. 10S, where the case is cited that four Old Slavonic verbs, jesmi, vemi, dami, jami, have effected that in New Slovenian and New Servian all the verbs of all classes of conjugation end in m in the first person singular number. In some forms of the intensive from tri-consonantal stems, where it is formed by the repetition of the third stem-consonant, we may have the influence of the analogy of originally bi-consonantal stems where such forms are organic. Such are Job 3. 18 'they are at ease,' I^P^ 'they languish.' Under this head may be placed also those plural forms of modern Syriac in which the stem of the singular is enlarged by the repetition of the final stem-consonant (Nöld. Neus. Sprache, p. 143, §72). These are probably, as Nöldeke remarks, based on

4°

AMERICAN

JOURNAL

OF

PHILOLOGY.

s u c h plural forms as 'amamln, H e b r e w ^ODIi. T h e instances of this k i n d are g i v e n b y N ö l d e k e loc. cit. A m o n g t h e m are (milAlg) ' w o r d s , ' N33"12 (birkäki) ' k n e e s , ' NSQ^B (tilpäpf)'eyelashes.' U n d e r this h e a d m a y also be p l a c e d those cases w h e r e a p h o netic process, starting u n d e r certain fixed conditions, was afterwards applied to stems w h e r e these Conditions did not exist. In s u c h cases the s t r o n g or r e g u l a r v e r b leaves its n o r m a l p a r a d i g m or t y p e a n d follows the a n a l o g y , not e x a c t l y o f a w e a k stem, but o f a s t e m w h o s e stem-consonants, a l t h o u g h s t r o n g a n d firm, still h a v e s o m e peculiarity which causes a c h a n g e in the inflection. T h e S e m i t i c v e r b forms its reflexive b y a p r e f i x e d it or ta. In E t h i o p i c it is ta. In H e b r e w the form of the reflexive prefix is nn. In stems b e g i n n i n g with a sibilant the n o f this reflexive prefix is t r a n s p o s e d s o as to c o m e after the sibilant instead of before it. T h u s , ' f r o m löB» «to k e e p , ' ' t o watch,' the imperfect reflexive is i B F f ? ' he will o b s e r v e (for h i m s e l f ) . ' T h i s m o d e of f o r m i n g the r e f l e x i v e from stems b e g i n n i n g with a sibilant b y m e a n s of an infixed instead of p r e f i x e d t is found in nearly all the S e m i t i c l a n g u a g e s . In A r a b i c and A s s y r i a n , h o w e v e r , this m o d e of f o r m i n g the r e f l e x i v e b y m e a n s of infixed t (ta or tau) is not confined to stems b e g i n n i n g with a sibilant, but the reflexive p a r t i c l e is universally infixed. 1 In this case the s t r o n g p r o b a bility is that it was not so in the primitive form of the l a n g u a g e , but that it started, as in H e b r e w , with the stems b e g i n n i n g with a sibilant, a n d then the influence of these formations g r a d u a l l y spread until it included all stems. O n l y t w o other explanations are p o s s i b l e ; either the primitive S e m i t i c form was indifferently p r e f i x e d or infixed t, or else it was universally infixed, and b e c a m e p r e f i x e d in H e b r e w , A r a m a i c , etc., e x c e p t in the case of s t e m s b e g i n n i n g with a sibilant. Praetorius, in his article " Ä t h i o p i s c h e G r a m m a t i k und E t y m o l o g i e , " in the first n u m b e r of the B e i t r ä g e zur A s s y r i o l o g i e , edited b y Delitzsch and H a u p t (p. 25 a n d note), s a y s that it is not u n l i k e l y that the r e f l e x i v e t w a s o r i g i n a l l y infixed in E t h i o p i c . T h e p r o o f of this w h i c h he cites, h o w e v e r , in the case of kaddna ' t e x i t ' s e e m s v e r y weak. 1 In A s s y r i a n the t is infixed not only in the reflexive derived from the P a l (iftdlix), but also in the reflexive derived from the intensive stem (uptallix). In the latter formation the t is prefixed in A r a b i c (V and V I ) , not infixed as in t h e eighth form.

ANALOGY

IN

THE SEMITIC

LANGUAGES.

41

D e l i t z s c h , on the o t h e r hand ( A s s y r i a n G r a m m a r , §83), r e m a r k s that the reflexive prefix was o r i g i n a l l y p r e f i x e d and not infixed, a n d that traces of this p r e f i x i n g are still found in A s s y r i a n . T h i s , too, favors the position that the c h a n g e from prefix to infix first t o o k place in the case o f stems b e g i n n i n g with a sibilant, to a v o i d c a c o p h o n y , and was then, b y the force o f a n a l o g y , e x t e n d e d to all stems. A similar case of the extension of a phonetic c h a n g e b e y o n d its o r g a n i c limits b y the process of a n a l o g y is found in A m h a r i c , in the linguistic p h e n o m e n o n k n o w n as ' p a l a t a l i z a t i o n ' 1 ( G e r m a n , M o u i l l i r u n g ) , in w h i c h the pronunciation of a consonant is affected b y a certain liquid modification occasioned sometimes b y a following i, j , or and s o m e t i m e s also t a k i n g place without the influence of these sounds. In the cases w h e r e the palatalization is occasioned b y the sounds mentioned, certain modifications in vocalization are phonetically legitimate. In s o m e cases, h o w ever, these c h a n g e s in vocalization take place w h e r e t h e y are not o r g a n i c , but analogical. T h e r e is no i, j, or ¿-sound to occasion t h e m , y e t the l a n g u a g e acts as t h o u g h there were, s i m p l y b e c a u s e t h e p h e n o m e n o n of palatalization is usually a c c o m p a n i e d b y these sounds. In the pronominal suffix of the 3d p e r s o n plural, masculine or feminine, atjaw, for instance, the liquid modification o f the t is not occasioned b y a following i,j, or i s o u n d . Y e t the l a n g u a g e acts as t h o u g h there must be a latent ¿-sound, a n d in s o m e cases places it there ; s o in the form bil&ij&w ' w h e n he told t h e m . ' S o also it often h a p p e n s that a stem which originally e n d e d in w or j lost its final stem-consonant in the m o d e r n A m h a r i c b y a p r o c e s s of p h o n e t i c d e c a y . T h e theme thus shortened is simply inflected as t h o u g h it never h a d another stem-consonant, or else a trace of the v a n i s h e d final j is left in the palatalization of the p r e c e d i n g stem-consonant. In s o m e cases, h o w e v e r , w h e r e this palatalization has t a k e n place without the influence of a final j the l a n g u a g e p r o c e e d s on the p r e s u m p t i o n that there must h a v e existed s u c h a final stem-consonant. T h u s the v e r b manatja never existed in the form manataja, y e t it forms its imperfect, not imantj, as it s h o u l d do, but imanatj, as t h o u g h there were a final stem-consonant/. S e e this w h o l e matter fully treated and illustrated in Praetorius, A m h a r i s c h e S p r a c h e , p. 45, §23b, and p p . 237-240, §200. 1

C f . Beitr. z. A s s y r . I 257, n. 9.

42

AMERICAN

JOURNAL

OF

PHILOLOGY.

CLASS II.

Analogical formations involving changes and confusion in vocalization. In Semitic we distinguish in the nominal formations between substantives of internalformation (Stade : Nomina innerer Bildung ; Barth : Schlichte Nomina) and substantives of external formation (N. äusserer Bildung). Substantives of external formation are formed either b y prefixes or by suffixes. Substantives formed directly from the stem without consonantal additions are divided into different classes according to differences in their vocalization. According to Barth, Neminalbildung, these different nominal formations in Semitic exhibit many instances of the analogical influence of one form or class of forms upon another in the matter of vocalization. T h e following is a brief and summary statement of the fundamental positions of his book : T h e nominal formations of Semitic are formed either from the perfect or from the imperfect verb-stems. There are three classes of perfect verb-stems, distinguished by three characteristic vowels, d, i, ü. Forms with a are transitive. F o r m s with i and u are intransitive. E a c h of these classes of perfect-stems gives rise to a corresponding class of nominal formations likewise distinguished by the three characteristic vowels, a for the transitive, and i and u for the intransitive. T h e r e are also three classes of imperfect stems, distinguished likewise by the three characteristic vowels u and i for the transitive, a for the intransitive. Each of these three classes, again, gives rise to a corresponding class of nominal formations. In many of these nominal formations we observe that the characteristic vowel of the class to which it belongs has been correctly maintained. These call for no further attention. In other cases, however, we find that a change has taken place. That is to say, the verbal stem, either perfect or imperfect, as it now exists, has a different vowel from that of the corresponding nominal formation. In cases such as these there are two possible explanations. Either, in the original form of the language the verbal stem existed in two forms with two different characteristic vowels, one of which subsequently ceased to be used, surviving, however, in the corresponding nominal formation ; or else the anomalous characteristic vowel of nominal formation is the result of analogy.

ANALOGY

IN

THE SEMITIC

LANGUAGES.

43

I have noted the following cases as cited by Barth : §17(3. Intransitive nominal forms of the form qutfill, legitimate when formed from «-perfects, but sometimes also found derived from z'-perfect stems. T h u s 'umudd (perf. z) 'full of youthful strength.' Sometimes intransitive adjectives of this form are formed from transitive stems. T h u s §umull 'hard, firm,' kubunn 'avaricious.' P. 36, §24. Intransitive nominal forms of the form q&tl derived from /-perfects. In A r a b i c : zuhd from zahida ' to be white, glossy.' husr ' l o s s ' from hasira ' to lose.' *ujb from 'ajiba ' to be astonished.' rukb from rahiba ' t o fear.' rugb 'avidity, voracity' from ragiba 'to wish, desire.' In Hebrew: IPX ' faithfulness,' from (?) ' to be faithful.' ' w a n t ' from ">pn ' t o want.' In S y r i a c : ' a n g e r ' from 'to be angry.' from ' t o be strong, heavy.' W W from l»!?'. Kani 'laughter' from ' t o laugh.' from 'to be whole.' KpTrn ' l e n g t h ' from p m Barth, p. 43, §28c. Instances of the form qatil intransitive from «-perfect stems. sari' ' rapid ' from saru'a 'to hasten.' jalld ' h a r d ' from jaluda ' t o be hard, firm.' karlm 'noble' from karuma. kabir

' old ' from

kabura.

kabira, fut. a, 'to be advanced in age,' is, however, also given in the dictionaries. katir ' m a n y ' from katura. 'asirn 'great, large, immense, h u g e ' from 'azuma. In the T a r g u m s : ' d r y , ' "pit? 'quiet,' 'sleeping,' TOT 'dark,' from stems having u in the perfect. T h e same phenomenon is observed in nominal derivatives from imperfect stems (see page 96, §64d,e). T h u s the Hebrew infinitive 7DP is formed not only from «-imperfect stems, where it is

44

AMERICAN

JOURNAL

OF

PHILOLOGY.

organic, but also from intransitive verbs with «-imperfect, e. g. 'to sleep,' 3ns; < t0 love,' 'to cease,' ^ 'to hate,' NT 'to fear.' These are clearly the result of analogy. In other cases, however, we must consider the possibility that in a previous period of the language the two forms with two different characteristic vowels existed side by side, the one having survived in the present form of the verb-stem, the other in the corresponding nominal formation. Thus in Hebrew imperfect a is the regular imperfect of stems having a guttural as second or third stem-consonant, while the corresponding nominal forms often point to an original u or i form. Such are WS 'to strike,' &P? 'to be angry,' n V ' to send.' So also transitive qitl forms in Arabic from verbs mediae gutturalis which now have only imperfect a,fi'l 'act,' sihn 'charm,' rilile 'passage, journey.' Here it is more probable that the uniform «-imperfects are themselves the result of a later modification under the influence of analogy. So also in the case of transitive nominal forms of the form qdftl or qittl derived from the imperfect stems. The imperfect i itself has been mostly crowded out by the prevailing analogy of the w-imperfects, while the original vowel has often been preserved in the corresponding nominal formation, especially the infinitive. (See Barth in Z D M G 43, 177-91.) The following instances are given. Infinitives with suffixes: D-J?D ("DO 'to sell'), n5B\ Infinitives without suffixes: l ' ^ Deut. 26. 12 'to tithe'; also the noun 'tenth'; "Hl^ 'to help' 2 Sam. 18. 3 (Kethibh), P^tf? 'to receive a portion.' Sometimes even when the «'-imperfect was preserved, and not crowded out by the prevailing analogy of the «-imperfect, its nature was misunderstood and this misunderstanding led to further analogical formations. In many cases the language mistakenly assumed them to be Hiphil-imperfects, with which they completely coincided in lorm, and then by analogy produced other Hiphil forms to correspond with them. So especially in the case of stems y'V. From D\ anj, etc., were formed the Hiphil-forms P^n, D'bp, vanp, p m Ttij. etc. The same thing took place in case of the other stems. According to Barth (op. cit. p. 119, §783) : Whenever a transi-

ANALOGY

IN

THE SEMITIC

LANGUAGES.

45

tive form qitl of any stem points to an original z-imperfect, and when this stem occurs in some languages in the Qal-form, but in others only in the Hiphil, the presumption is that this Hiphil is of secondary formation, based on a misconception of the original ¿-imperfect. Thus the common Semitic form Arabic dikr, Ass. sikru points to an original z'-imperfect stem jazkir\ compare Aramaic with a on account of 1. As, however, this z-imperfect has disappeared from the Qal in Hebrew, it is probable that it was retained as a Hiphil, and then the other Hiphil forms were formed as a later analogical development. Arab, liadana 'to take in the arms' has at present only the «-imperfect. From this we have Hebrew 'arm, bosom,' and Arab, hudne 'embrace.' But the Hebrew 13VD and Arab. hidn point to an original z-imperfect. From ^afl 'reverse, contrary,' and npsq 'overthrow,' we infer the former existence of a form ISP! as ¿'«/-imperfect. This is not found now in north-Semitic. But Syriac has Aphel in the same sense as Hebrew and Syriac Qal. Compare also Hebrew •qsnn Job 30. 15, which presupposes the Hiphil. In the same manner the noun siqi (watering) and the imperfect iasqi explain the formation of the later Hiphil, ^""h and Aram. W K , resulting in crowding out the Qal which must have existed at a former period ; cf. n ^ , Syriac etc. So also the Hebrew TPJ 'libation,' form qatll, Deut. 32. 28, points to an original z-imperfect. Hence Ps. 16. 4, Gen. 35. 14 are regarded by Barth as originally z'-imperfects Qal. Being erroneously regarded as Hiphil-imperfects they furnished the basis for further analogical Hiphil-formations. Another case is Being regarded as a Hiphil-imperfect it furnished the basis for further analogical Hiphil-formations, but n^np, D^ni?»- and ni^ripp point to an original z-imperfect of the Qal. P. 116, §77a, 3. In Arabic the transitive form qitle is formed organically from z'-imperfect verb-stems to express the mode or manner of the action expressed by the verb. After this meaning had become established for this form it was extended by the force of analogy to verbs of every class. P. 126, §82. Transitive nominal formations qut&l, qdtiil formed from the imperfect. W h e r e such formations exist from stems which at present do not have u as the characteristic vowel of the imperfect we must

46

AMERICAN

JOURNAL

OF

PHILOLOGY.

again, as elsewhere, distinguish t w o c a s e s ; either at an earlier period the « - i m p e r f e c t existed side b y side with an i- or «-imperfect, a n d w a s g r a d u a l l y displaced, or else the « - n o m i n a l formations are the result of a n a l o g y . Instances of the f o r m e r case a r e seen w h e r e the A r a b i c has o n l y z-imperfect while H e b r e w a n d S y r i a c h a v e « - i m p e r f e c t ; so A r a b i c hulUk; ^ H ! and guftir. But m a n y s u c h cases are p r o b a b l y the effect of a n a l o g y . T h e following are instances of infinitives of this class w h e r e there existed no c o r r e s p o n d i n g « - i m p e r f e c t : uutjib from iiataba ' t o leap.' •uurtid f r o m uarada ' t o arrive.' uuj&d from uajada ' t o find.' uujiib from uajaba ' t o be necessary.' y.u'ill from ua'ala ' t o s e e k r e f u g e in.' itujuz from uajaza ' to be brief.' P- x 3 5 . §85^. T h e infinitive qatll is used t o form collectives. O r g a n i c a l l y it b e l o n g s , o f course, to «'-imperfects, but it is often f o u n d w h e r e there is no c o r r e s p o n d i n g ¿-imperfect, especially to d e n o t e collectives of the n a m e s o f animals. S o kalib ' d o g s , ' 4a'in ' s h e e p , ' mdiz ' g o a t s , ' naqtd'cattle.' e. In E t h i o p i c the infinitive qatil has b e c o m e the r e g u l a r v e r b a l infinitive for e v e r y class of verbs, a n d as s u c h is f o r m e d from all the different c o n j u g a t i o n s . T h i s analogical extension o f the infinitive form qatil has c o m p l e t e l y c r o w d e d out the old parallel form qdtil. f . S o also in later H e b r e w ( M i s h n a ) the feminine n^'Dp, u s e d to denote abstract notions is f o r m e d without r e g a r d to the character o f the stem. T h e instances g i v e n are, iiD'On, npna, n^SJ, ns'X», m»DX, n j v m na^n, P. 144, §92. Transitive qatilat. In the case of the w o r d s nap;). a n d '"tsip w e a r e unable t o determine w h e t h e r w e h a v e the e v i d e n c e of a former z'-imperfect o f these stems, or analogical .formations induced b y the signification of the s a m e form from other stems. P. 174, §122c. S p e a k i n g of the active participial form qdtul (or q&ticP) he s a y s that a g e n e t i c a l connection with the infinitive qutiil is b e y o n d question, and b o t h g o b a c k in their origin to the w-imperfect. B u t j u s t as it was s h o w n (in p a r a g r a p h 82 B of this s a m e w o r k ) that the infinitive qutM was often f o r m e d analogically w h e n there a p p e a r s no «-imperfect, so also the participial form qdtM is u s e d to f o r m an active participle from a n u m b e r of stems w h i c h s h o w o n l y the ¿-imperfect; so dardb ( b y the side of darib)

ANALOGY

IN

THE SEMITIC

LANGUAGES.

47

'striking,' hattif 'buzzing, sounding, clanging,' gasAm 'violent' (but gasama is given with «-imperfect), kasiib 'gaining, acquiring.' P. 175, §i22a, note 1. Instances of active participles of the form qatM formed from stems having z-perfects, through the influence of analogy. In Arabic: dahtik 'laughing,' 'amUl 'doing,' ia'-ds ' despairing.' In Hebrew: E>SJ 'KJEf 2 Sam. 5. 8 and i?l rains? Hosea 3. 1, are cited as instances of this form. From the Mishna, pplDttn 'busying themselves,' Para 4. 4, nsitojn rvt 'dripping olive,' Pea 7. 1 llpc? 'in 'be watchful,'Ab. 2. 14 'dwelling,' etc. P. 178, §i23a. T h e form qattil as passive participle is formed in Hebrew from all stems indiscriminately. P. 186, §i26a. The participial form qatil organically connected with z-imperfects is formed in Arabic from verbal stems of every kind. d. In Aramaic this form has become the regular passive participle, and as such is formed from every transitive verb. P. 201, §i36b. Participle q&til. This form arises from the transitive z-imperfect. The form from the z-imperfect, rather than that formed from the «-imperfect, became the prevailing one through the influence of the forms of the participle in the derived conjugations, which show an z-vowel after the second stem-consonant. T h e only difference between the participle of the Qal and that of the derived conjugations lies in the fact that the latter prefix an m, while the former, not taking the prefix, lengthen the vowel of the first syllable. After this form was established as the participle of the transitive z'-imperfect, it was extended to all transitive imperfect stems, and also to those stems which are intransitive in structure, but transitive in meaning. In a few cases it is extended to verb-stems intransitive in meaning. Arabic, timin 'sure,' s&lim 'safe,' b&'id'iax.' Eth. rate 'true.' Aram, ^EH 'fearing,' «10 'rejoicing.' Heb. n^lT 'base, abject.' I have stated these principles and given these illustrations just as they are given by Barth. I must say, however, that to my mind his whole position and the value of all his discussions are

48

AMERICAN

JOURNAL

OF

PHILOLOGY.

e x t r e m e l y doubtful. F o r , first, as he himself admits in m a n y places, w e d o not k n o w h o w far these forms, which a p p a r e n t l y s h o w a different v o w e l from that of the form from w h i c h he derives them under the influence of a n a l o g y , m a y g o b a c k to other forms which no l o n g e r exist, b u t w h i c h , w h e n existing, had the same v o w e l as the nominal formation d e r i v e d from them. T h e w h o l e of §100, for instance, on the imperfect infinitives of the d e r i v e d c o n j u g a t i o n is m a d e e x t r e m e l y uncertain b y the admission w h i c h he r i g h t l y makes, that formerly these imperfects m a y have s h o w n an u as well as an «-vowel. T h i s but s h o w s how uncertain the w h o l e matter is, far too uncertain, in fact, to form the basis of scientific investigations and conclusions. B u t again, I d o not see that he establishes with a n y d e g r e e of certainty the fundamental position of his w h o l e b o o k , viz. that all the nominal formations must be referred either to the perfect or imperfect verb-stems, w h o s e characteristic v o w e l will n o r m a l l y be the characteristic v o w e l of the c o r r e s p o n d i n g nominal formation, and that all deviations therefrom are the result of a n a l o g y . Why s h o u l d all nominal formations necessarily c o m e t h r o u g h the channels of the perfect and imperfect v e r b - s t e m s ? W h y can they not be referred directly to the simple g r o u n d - s t e m , without the intervention of the perfect or imperfect verb-stems ? T h e s e are questions not satisfactorily answered b y Barth, while t h e y are s u g g e s t e d b y the m a n y deviations from the principle w h i c h he s e e k s to establish. ABEL H.

HUIZINGA.

AMERICAN

JOURNAL

OF

PHILOLOGY

V O L . X I I , 2.

I.—ANALOGY

WHOLE

IN T H E

SEMITIC

NO. 46.

LANGUAGES.

PART III.

Besides such analogical formations as have just been presented and discussed, there are also many such formations in the inflection of the substantives. In the various processes of inflection, such as the formation of the construct, emphatic, plural, and in the addition of pronominal suffixes, each class is governed by its own peculiar laws. In a great many cases the law of the group maintains itself and no confusion occurs. But from the nature of the case, considering the subtle nature of these formations and distinctions, we might expect occasional confusion and transitions from one class or type to another, in other words, metaplasms and analogical formations. A n d such, on examination, we find to be the case. Thus, according to Stade (§191), in Hebrew the nominal forms qatl, qitl, qutl form their plurals after the analogy of the forms qatal, qital, qutal On these plural formations last mentioned, however, we have an interesting dissertation by Mr. W . H. Salter Brooks, " Vestiges of the Broken Plural in Hebrew " (Dublin, 1883). The object of this paper is to show that the stems qatl, qitl, qutl formed their plurals originally without any plural termination whatever, but simply by internal vowel changes, thus: qatal, which became qatal, qital or qital, qutal or qutäl, exactly like the corresponding Arabic internal or broken plurals. According to this view, then, the plural of these words is sufficiently indicated by these internal 1 Cf. Haupt, Beiträge zur assyr. Lautlehre (Göttingen, 1883), p. 89, n. 3 ; Praetorius in Beiträge zur Assyriologie, Vol. I (Leipzig, 1890), pp. 374—7.

VOL. XII.—2.

134

AMERICAN

JOURNAL

OF

PHILOLOGY.

vowel changes. But subsequently, owing to the tendency to vowel corruption and obscuration, these distinctions were not sharp enough and clear enough to distinguish between the singular and plural. T h u s at first the Hebrew would have presented the scheme: s. bcCl.

pi. ba'al

s. rumh.

pi. rumah,

' lord.' 'spear.'

But owing to the presence of the guttural: consonant, the singulars were bound to become ba'al, rumah, and actually did so, that is, become identical with the plurals. Hence, in such cases, the addition of the external sign of the plural was necessary in order to distinguish the singular from the plural. But after it was added to such forms, the language lost sight of the distinction between singular and plural sufficiently maintained by vowel changes, and gave to the external or sound plural formation universal extension. T h e application of the external plural sign in these cases, therefore, is not to be regarded as organic, but as the result of analogy. O n this view of the matter we have here an instance of analogical formations Class III, D , the addition of formative elements where they d o not belong. In Aramaic and Syriac substantives of the type qatl sometimes form their absolute and construct state after the analogy of the type qatil. T h e organic form reappears in the emphatic state. ( S e e Haupt in Hebraica, Vol. I, p. 228, note i.) 1 T h u s in Syriac we have the regular normal forms in (for tar a) Ninn ' door.' ( f o r p a g a r ) N^B < corpse.' But ' s o u l ' stands for napis.(form qatil). T h e emphatic is syncopated from napv>&. S o also KS?? 'shoulder,' and 'king,' etc. O n the other hand, the ground-form of 'lord,' is ba'l (type qatl), construct ba'al, and from this we ought to have, by the laws of Syriac phonology, the form , and the /-vowel of the second syllable is due to the analogy of forms like in other words, a metaplasm from the type qatl to the type qatil. S o also D1?< Kp-i? 'stomach,' K»1?* 'image,' 'reason,' etc. 1 Cf. Lagarde, Bildung der Nomina, pp. 73 and 74, also Beitr. z. Assyriologie, I 325 and 375.

ANALOGY

IN THE SEMITIC

LANGUAGES.

135

Again, all vowel changes in the stems of words are governed by regular phonetic laws. (See, e. g. the laws o f Hebrew vowel changes in Stade, §73 fg.). But in individual cases analogy often plays an important part. In such cases an individual is transferred from one group to another, or a phonetic process is taken out of its legitimate sphere and applied to cases where it is not legitimate. An example is given in Stade, §78b. A long 'arrange,' 'collect,' we see an original i-vowel. Not so, however, in the 2 pers. masc. pi. Here nearly all the forms which originally had the characteristic vowel u now have the vowel i. The change has been made under the influence of the analogy of those forms where the z-vowel is original. Only three cases remain in which the original «-vowel has been preserved, viz. •1310 from 'to be laid waste.' from I r a 'to draw.' •'HI?' from "nt? 1 to oppress, destroy.' For instances of forms of the Assyrian verb where the force of analogy has resulted in changes in vocalization, see Haupt, " Der Keilinschriftliche Sintfluthbericht," in K A T . 2 68, note 2.1 By the side of the older organic forms of the present, such as ib&lut 'he lives,' isagum ' he cries,' iz&nun ' it rains,' we have later analogical formations, such as izanan, isakan ' he places,' where the vowel of the last syllable has been influenced by the vowel of the preceding syllable. Delitzsch, however (Assyrian Grammar, §9oa, note, also §94, and §115), maintains the contrary view, that izanan is the organic and izanun is the analogical. 2 Finally, on comparing the vocalization of the Hebrew intensive perfect and imperfect with the corresponding forms in Arabic, it seems probable that the Hebrew vocalization has been changed under the influence of analogy. In Arabic qattala, yuqattilu is probably nearer the original in form, and more organic, while in Hebrew the vowel of the second syllable has been influenced by the imperfect. Compare Barth, "Nominalbildung," p. xxii fg., also p. 70. On this point Dr. Martin Schultze (Zur Formenlehre des semitischen Verbums, Wien, 1886, p. 48, §49), remarks that the derived stems are undoubtedly younger than the simple stem or 1 5

C f . Beiträge zur Assyriologie, Vol. I, p. 124. C f . Barth in Z A I I 3 7 7 ; Z i m m e r n , Z A V 10.

ANALOGY

IN

THE SEMITIC

LANGUAGES.

139

Qal. It is to be assumed that they first formed the younger tense (the so-called imperfect), whose vocalization is the same in all the different dialects. But at a subsequent time, when the need was felt to form the afformative tense (or perfect), the tribes had separated. The older northern Semites retained, without reflection, the vowel of the future; the younger Arabs, however, chose, more consistently, the vowel of the Qal-Praeterite. CLASS III.

Analogical formations with reference to the formative elements of words. Division

A.

Mistaking a servile or formative element for part of the stem. An interesting case of this kind is described by Stade, Hebrew Grammar, §400-2. It is in the formation of the 3d person feminine singular perfect of the n'6 verbs. The ground-form was qataiat (compare Arabic radiiaf). This could become either qatM or q&tdjd. The former has been preserved in new Lev. 25. 21, stem HOT 'to make.' 2 Kings 9. 37 Ketkib, stem n»n 'to be,' and with suffix new. The latter is preserved in n^Dn p s . 57. 2 '(my soul) has fled (for refuge).' But the form qdtdi was generally regarded as a regular perfect tri-consonantal stem ; the servile n was erroneously regarded as a stem-consonant. Then, taking the forms qalela and the pauseform qtttlt as models, the feminine ending d was appended, giving the form qMAt&. This form has been retained in the pause, n n w . nrpn, etc. From this, after the analogy of qatel&, we have the usual form n Jvn. Another case of this kind is seen in the treatment of the Niphal of the l>"l?-stems. The ground-form is n&katt. The organic form is seen in stem Tin 'to be dry.' SD3, Stem 330, ' to turn.' stem 'to be light.' But in other cases it is written as though it were the Qal of a verb J"S, the } preformative being erroneously regarded as a stem-consonant. So in the pointing 2D3. DDJj stem DDDJ 'to melt.' P i Eccl. 12. 6, stem 'to break.'

I40

AMERICAN

JOURNAL

OF

PHILOLOGY.

Then, as from 13? we have in the feminine so we have the feminines stem pp3, 'to empty.' and the plural , stem 'to shake.' Sometimes a n, which really is the sign of the feminine, is treated as if it were a stem-consonant. T h e examples of this class of analogical formations found in Syriac are given by Nöldeke, Syriac Grammar, §78 and 80. The same thing has taken place in Ethiopia See Dillman's Grammar, §133, and §137, 5b. For the Assyrian examples of this class see Haupt, S F G M 73;' also Delitzsch, Assyrian Grammar, §69. muttäti, pi. of muttu. Mt&t, pi. of letu 'night' (st. isitäte, pi. of isitu ' pillar.' 2 On the Hebrew nirfci, mnrp, etc., consult A m . Journ. Phil. X I I 37; also Barth, in Z D M G 41,605-607, and Nominalbildung, p. 4b. In Syriac the formations n i - xrn- from nouns in Kn-1-, are organic only in derivatives from stems so KJn-lD1!, pi. xnif?^ ' image.' In forms like K^I^D 'kingdom,' the ending n-1, which is merely a formative, is treated after the analogy of those words where it forms part of the stem. It might be expected that languages, starting out, as do the ancient Semitic, with a sharp distinction between stem-consonant and servile or formative element, would in the course of time and a development extending over more than a thousand years, lose the sharpness of this distinction and incorporate elements which originally were formative as part of the stem. This expectation is borne out in the study of the Amharic, the Mandean, and the modern Syriac. See Dr. Martin Schultze, " Zur Formenlehre des semitischen Verbs," p. 43, §45. Thus in Praetorius, "Amharische Sprache," pp. 130-140, under the title " Wurzelerweiterung," we have a study of those stems of the Amharic language which have been extended beyond the original three stem-consonants, mainly by incorporating and reckoning as part of the stem those consonants which originally were formative. In many cases, however, as is pointed out by Praetorius, the 1

C f . , h o w e v e r , H a u p t , B e i t r ä g e zur A s s y r i o l o g i e , V o l . I , p. 16.

2

C f . L ö t z , T i g l a t h p i l e s e r , p p . n o , 56 a n d 1 5 5 , 27.

ANALOGY

IN

THE SEMITIC

LANGUAGES.

141

incorporation of the formative element is preceded by the loss of one or more of the original stem-consonants, so that the formative element is taken as a compensation for this loss, and the appearance of having three stem-consonants is restored. The other methods of extending the original stem, viz. by doubling the original bi-consonantal stem, and by other additions, or by a repetition of one of the original stem-consonants of the tri-consonantal stem, do not, of course, come up for consideration in this paper. W e are concerned only with those cases mentioned by Praetorius as constituting the third method of extending the stem, namely, by incorporating a nominal, pronominal or verbal formative element, or even a preposition, as part of the stem. I add a list of the cases mentioned by Praetorius, referring to his work, however, for their fuller discussion. The causative prefix a is treated as a stem-consonant in the stems amara as if it were a simple stem primae gutturalis ; but in reality it has arisen from 'amara, the causative of a stem med. gutt. So also 'adaga ' to grow.' 'asalama ' to be a Mohammedan.' A more doubtful case is 'arasa ' to moisten.' The causative prefix sa or its modification sa is treated as a stem consonant in the following words : sanakala ' to give offence, to be a stumbling-block' (Anstoss erregen). shanagala 'to deceive.' shamagala 'to become old.' shanakara 'to leak.' shamatata ' t o despise, scorn' (perhaps). shamana ' to weave.' shaqaba 'to ascend.' This seems to have been developed from 'ashaqaba 'to lift up,' a second causative from aqaba. So in the " Quinqueliterum " shakarakara 'to rise,' and sabaraqa 'to shine, sparkle.' The following stems have been formed by raising the t of the reflexive prefix ta to the dignity of a stem-consonant: tafia, ' to sleep,' from tanahaya ' to rest oneself, to recover.' tanafasa ' to breathe, to take breath.' ianakuala ' to be deceitful.' talala ' to be deceived.'

142 talalafa

AMERICAN

JOURNAL

OF

PHILOLOGY.

' t o cross.'

tawaba ' to be beautiful.' tarasa forms the new double reflexive taniarasa 'to support one's head by anything.' tanabaya 'to prophesy.' Perhaps also takata 'to be slow, lazy.' Sometimes the reflexive t has been weakened to d, and is then treated as a stem-consonant. So in dasasa, 'to feel of.' Perhaps also daraqa 'to be dry' belongs here, and danaquara 'to be deaf, senseless.' danagara 'to dissuade.' Here also Praetorius mentions the Ethiopic danagada 'to frighten,' and danagala. Nominal formative elements are likewise incorporated as part of the stem ; so m in the words mdshara 'to be a bridegroom,' from Ethiopic moshar ' a saw.' mashaga 'to barricade oneself.' marakuaza

' t o support.'

masaqala manazara manatsha managaga m&kuata

'to hang.' 'to commit adultery.' 'to spring forth.' 'to hold,' or 'to open another's mouth.'

' t o vow.'

mclraka 'to take captive' (also in Ethiopic). mdsana 'to recommend' (also in Ethiopic). So the prefix 'a of internal plural formations is reckoned as part of the stem in 'amalaka 'to adore God.' 'asam& 'to practise witchcraft,' formed from the plural 'asm&t 'witchcraft' with the loss of the t. So also the nominal suffix t in the words gazata ' to excommunicate,' from the Ethiopic stem wagaza. So perhaps from a noun with the suffix ut or dt we have tanfawata 'to offer opposition, to withstand.' galamSta 'to play the whore,' from galamoi 'whore.' So the consonant n of the abstract ending in shalatana ' to rule.' m and t are treated as stem-consonants in malakala 'to denote,' from lak'a. mdgata 'to dispute,' from wag'a 'to push, to strike.' mtxlada ' t o get up early.'

ANALOGY

IN

THE

SEMITIC

LANGUAGES.

143

S o the consonant m of the adjective ending dm is reckoned as part of the stem in such words as quaratama ' to have the rheumatism ' and taraqama ' to fill up, to stuff' (vollstopfen) and garamama ' t o exercise a horse.' Other noteworthy cases are the following : zawatara ' to continue, g o on,' in which the relative za has become part of the stem. baratä 'to be firm, strong,' in which the preposition ba has become part of the stem. Praetorius further mentions similar formations in Ethiopic, such as la'dla ' to be high,' from la+ *?!!. A l s o bahata and bähtawa ' to be alone,' from ba + 'ahati. F o r a very complicated case of this kind see Praetorius, Amharische Sprache, §2o8a, b, c, d. 1 In an early stage of the language the verbal infinitive of heda ' t o g o ' united with the auxiliary verb 'ala ' t o b e ' and the union became fixed both in form and in meaning. A t a later period, b y a "false a n a l o g y " this form, the result of this union, was considered a normal homogeneous verbal stem, and received further modifications, so that its origin was much obscured. T h e verbal inf. of heda, which at present, in accordance with the regular process of contraction in Amharic, is hldd, was formerly haydd. This older form haydö united with a following 'ala, and formed, not haydwäl, which the later phonetic law would require, but haydöla, in accordance with an older law still prevalent in Tigrifia (see Praet. Tigrina Grammar, §48, p. 69), and which accordingly we must postulate for the older period of the A m haric. haydöla then m e a n t ' he has gone, he has left, he is not here.' But in course of time the original meaning of the verbal stem ' t o g o ' was lost sight of, and the negative element, from being merely secondary, came t o b e regarded as the primary sense ; and so the initial hay was regarded as a sort of negative particle. Then, in accordance with the analogy of the initial syllable of the negative form of the imperfect, hay became 'ay, and finally the negative m was attached, 'aydölam ' he is not.' Since, then, the first part of this form, viz. 'aydd, had lost its original character as a verbal infinitive, it had to lose at the same 1

See also Praetorius' B e i t r ä g e zur äthiopischen G r a m m a t i k und E t y m o l o g i e ,

in V o l . I of B e i t r ä g e zur A s s y r i o l o g i e , e s p e c i a l l y § § 1 , 2, 6, 52.

144

AMERICAN

JOURNAL

OF

PHILOLOGY.

time the ability to m a k e its suffix c o r r e s p o n d to its subject, and so it b e c a m e f i x e d in the form of the 3d pers. masculine singular. T h e form 'ayddlam h a d the a p p e a r a n c e of a n e g a t i v e imperfect of a stem dola, o n l y the final a g a v e it the character of a perfect form. A n d since the e n d i n g s of the perfect w e r e a s s u m e d for all the different persons and numbers, the form of the 3d pers. pi. 'ayddlum, a l t h o u g h h a v i n g e x a c t l y the a p p e a r a n c e of a n e g a tive imperfect, w a s not able to transfer the w h o l e inflection to the imperfect, e v e n t h o u g h s u c h a transfer w a s favored b y the c h a n g e of the initial hay t o 'ay. A l s o the circumstance that the form with the relative is y ay dola ' w h o is not,' and not yamaydola, shows clearly that the form was r e g a r d e d as a perfect. In this w a y it c a m e to pass that this old c o m p o u n d verbal infinitive was inflected in a m o n g r e l w a y , in its initial part h a v i n g the. appearance of a n e g a t i v e 3d pers. imperft., a n d in its final part h a v i n g the a p p e a r a n c e of a perfect. T h e s e m o n g r e l f o r m s a g a i n h a v e resulted in further analogical formations. B e i n g r e g a r d e d as pure imperfects 3d pers. m a s c . sing, t h e y r e c e i v e d the usual p r o n o m i n a l object suffixes, m e a n i n g : ' it d o e s not b e l o n g to us,' or ' y o u or thee it is not a p p o i n t e d to,' etc. T h e n from this the reflexivetaddla ' to be appointed, assigned,' w a s f o r m e d , a n d e v e n the simple dola imposuit s e e m s to h a v e been used, t h o u g h Praetorius s a y s he never found it in any text. Similar cases are found in T i g r i n a . S e e Praetorius, T i g r i n a S p r a c h e , §173. T h e reflexive prefix ta is r e g a r d e d as part of the stem, a n d from the stem thus a u g m e n t e d a new c a u s a t i v e is f o r m e d b y pref i x i n g 'a. S e e the e x a m p l e s g i v e n in Praetorius, T i g r i n a S p r a c h e , §173. T h e M a n d e a n instances of this k i n d of analogical formations are g i v e n in N o l d e k e ' s G r a m m a r , p. 84, §74, p. 98, §87. The A r a b i c form asnata, d e n o m i n a t i v e from sane, sanat, ' e v i l y e a r , ' in w h i c h the feminine e n d i n g t is r e g a r d e d as part of the stem. In m o d e r n S y r i a c m a n y v e r b s with 1 as a fourth stem-consonant are denominatives from nouns which h a v e been f o r m e d b y a d d i n g the suffix 6in t o a tri-consonantal stem. T h e instances are g i v e n in N o l d . , N e u s y r i s c h e S p r a c h e , p. 192, §95. In other quadriliterals the first stem-consonant was originally a formative element w h i c h a f t e r w a r d s c a m e to b e treated as part of the stem. Such a r e m a n y b e g i n n i n g with and D, s u c h as ei'priD ' t o c h a n g e , ' a n d -onD ' t o visit,' e t c . ; others b e g i n n i n g with n, such as n m n

ANALOGY

IN

THE

SEMITIC

LANGUAGES.

I45

' to be boiled in p i e c e s o t h e r s beginning with JJ, such as D^SD 'to take out.' See the cases cited in Nold. op. cit. p. 194, §96. Some of these, formed from verbs with weak stem-consonants, have the appearance of original tri-consonantal stems. So the causatives from stems and ^'2 and V'l?; also from stems mediae D. See examples given in Nold. op. cit. p. 228, §108; p. 235, §110. Also p. 243, §113 ; p. 245, §113. Finally, we have under this class the somewhat analogous case where, not the formative element, but a consonant which is the result of a phonetic modification is treated as part of the original stem. See an illustration given by Haupt in Hebraica I 226. The Assyrian ittu 'side' is the feminine of idu 'hand,' and stands for idtu. The plural of idtu is idati, but sometimes it is formed from ittu, and gives us itdti. So also the construct aran 'sin,' from arnu, which is a phonetic modification of annu (resolution of the doubling by insertion of 1) is such an analogical formation. See Hebraica I, p. 219, note i. 1 CLASS III.

Division B. Mistaking a stem-consonant for a servile or formative element. W e find examples of this kind in Syriac in words in which the stem-consonant n is regarded as the sign of the feminine. See the examples cited in Noldeke, Syriac Grammar, §86. Also in §71, 11 KJ3B1K TFXVITK and K^app irraxrj, the 3 of the ending is a stem-consonant, but is treated as though it were a formative element. The Ethiopic behtew, solus solitarius, becomes in Tigrina behtti. The final u, however, was misunderstood and regarded as the suffix of the 3d pers. sing. masc. and the word was thought to mean ' he alone.' Hence the other corresponding suffixes were added to denote the other persons, genders and numbers. behtd 'she alone,' etc. (Praetorius, Tigrina Sprache, §121, 37). CLASS III.

Division C. Analogical changes in the formative elements themselves. Influence of one formative element upon another. 1

Cf., however, Zimmern, Babyl. Busspsalmen (Leipzig, 1885), p. 12, 6.

146

AMERICAN

JOURNAL

OF

PHILOLOGY.

The analogical changes which take place in the afformatives of the perfect in the inflection of the Semitic verb have been so thoroughly set forth by Noldeke in an article entitled " D i e Endungen des Perfects," in Z D M G 38, 407 fg., that the changes and formations need only be briefly noticed here, referring to that article for a discussion of the whole question. In Syriac we have the 3d person plural ending -tin for -ti of the other languages, after the analogy of the other plural endings. S o also in Assyrian we have in the permansive for kasdti sometimes kasdtini. In Arabic the 3d pers. plural feminine qatalna, in place of the common Semitic form qatald, shows the influence of the corresponding form of the imperfect, iaqtulna (see Dr. Martin Schultze, " Z u r Formenlehre des sem. Verbs," p. 18, §15). In the 2d person singular masculine the Ethiopic has -ka for the common Semitic form -t&. Noldeke explains this as due to the influence of the analogy of the possessive and object suffix of the same person. It is better, however, to regard it as the influence of the original ending of the 1st person sing. -kti, which has been preserved in Ethiopic, and which in that language has changed the endings of the 2d person masculine and feminine from -id and -ti to -ka and -ki, while in the other languages the reverse has taken place, viz. -kti of the first person was changed to -tti under the influence of the analogy of the 2d person, while in Hebrew a further analogical change took place, tti being changed to ti under the influence of the possessive suffix i. (See Haupt, S F G 52, note 10; Stade, {5179a ; Hommel, Semiten, 443, note.) In the 3d person plural masculine the Syriac forms in -d before suffixes are formed after the analogy of the imperfect forms in -tin and -on, which retain their original -a before suffixes. In the 2d person plural feminine the vowel -u of the Arabic ending -tunna is probably after the analogy of the same vowel in the corresponding masculine ending. S o in Ethiopic the pronominal suffix 3d pi. fem. hdn is probably after the analogy of the masculine hdmu. In the 1st pers. plural the Hebrew -nu for the common Semitic ending -nd is probably due to the influence of -nti in HOD as pronominal suffix is a late analogical formation and is found only with prepositions nspa. ¡"^n» (Stade, §348). So Dnj'p Gen. 1. 21 is based on the analogy of -inj'D.

ANALOGY

IN

THE

SEMITIC

LANGUAGES.

147

A s to the analogical changes in the preformative of the imperfect Qal I shall not attempt to discuss them here. See some remarks on this subject by Haupt in Beiträge zur Assyriologie, Vol. I, p. 17, note 20; also pp. 260, n. 27, and 328. Other instances where one formative element has influenced another are the following: T h e vowel of the preformative of the Hiphil participle in D'pD (stem Dip) and 3PP (stem 23D) is influenced by the vowel of the preformative of the perfect D»pn and 3pn. According to Stade ( § i 6 i b ) the n of the preformative of the Hebrew reflexive in as compared with the more primitive form 'it, is due to the influence of the "'> preformative of the causative. But the question then arises : to what is the n preformative of the causative due, when in Arabic we have äqtala and in Ethiopic 'agbdra ? The form of the ending of the imperfect plural feminine of verbs and V'l? has probably been influenced by the corresponding forms of the n"1? verbs. Thus nrapPi (stem 31D) after the analogy of nj^Fl (stem H51? 'to g o up.') The organic form appears in (stem 'to turn.') T h e form of the preformative of the perfect Hophal of verbs and has been influenced by the analogy of the same preformatives of verbs l''Q. •P_in) stem Dip, and stem 'to pierce, to begin' after the analogy of Tijin, stem ">">'', for i n 'to go down.' A case of this class is cited by Dr. Rosenberg in " D a s aramäische Verbum im babylonischen Talmud," p. 40. In the Ethpeel of verbs '"S, the ' loses its consonantal power in most of the forms, and consequently the A of the preformative remains. But in cases where the ' retains its consonantal power the n ought not to remain. If it does remain, it is due to the influence of the analogy of those cases where the ' loses its consonantal power. In the Ethpaal the organic formation is usually found. Another case where the form of one preformative has analogically changed the form of another preformative is the following, taken from Barth, Nominalbildung. T h e infinitive absolute (Barth, perfect infinitive) of the Niphal appears in Hebrew in three forms : Vtapj, "pb^n, and ^¡sn. O f these the first is the normal

148

AMERICAN

JOURNAL

OF

PHILOLOGY.

form. The other two are explained by Barth (Nominalbildung, p. 72, §49b) as analogical formations based on the analogy of the form of the infinitive absolute in the other derived conjugations, especially in the Piel and Hiphil. In these conjugations there was a certain similarity in sound between the infinitive absolute and the imperfect very noticeable when the infinitive was used to strengthen the imperfect; cf. Ex. 23, 24 ">3^ Hence this infinitive was supposed to have been formed from the imperfect, when in reality it was formed from the perfect. In the Niphal no such resemblance existed between the imperfect and the infinitive as formed from the perfect, and hence a new form was formed from the imperfect which should have the same assonance to the imperfect as was observed in the case of the infinitive absolute of the Piel and Hiphil. The ground of this explanation is strengthened by the fact that this analogical formation is found only when the infinitive absolute is used to strengthen the imperfect. W h e r e it is used to strengthen the perfect the normal form is used ; cf. on the one hand on the other hand Gen. 3i- 30. In forms like spxn the vowel of the last syllable is due to the influence of the old perfect infinitive of the Qal and Niphal. sp??, while in nnsn the assonance to the imperfect has been made complete; cf. tS^Bn, x Sam. 27. 1, 'escaping, I shall escape.' CLASS I I I .

Division

D.

Inorganic application of formative elements under the influence of analogy. The true origin, nature or significance of a formative element was often misconceived, and so it came to be attached to forms where it was not organically legitimate. A n instance of this class is given by Noldeke, Syriac Grammar, §5OB: " T h e 1 which was often found apparently without any special reason, came finally to be attached to words ending in a consonant"; cf. Hebraica II 104, n. 1. W e have another instance in the use of the old accusative singular ending & ¡"1- (accusative of direction) with 3 and IP, and in cases where it is attached to dual and plural forms, thus : n^»1? 'upwards,' HE!?1? 'downward.'

ANALOGY

IN

THE SEMITIC

LANGUAGES.

149

n o ^ f S 'to Chaldea.' 'to Heaven.' nonyp 'to Egypt.' n b a p 'from Babylon.' Sometimes there is a combination of two processes B and D of this class. T h e language mistakes a stem-consonant in a certain form for a formative element and then adds it where it does not belong. S o in the forms of Syriac numerals with determining suffixes (Noldeke, §149). T h e , which there appears is organic only in l . ^ ' we two.' After this analogy we have the other forms H ' T ^ f ' they three,' etc. In Ethiopic the pronominal form 3d pers. masc. sing, we'etu, forms the 3d person plural masc. and fem. we etomti and weetdn by the inorganic attachment of the suffixes -dmii and -on (Praetorius, Eth. Gramm. §21). Another case of the same kind in Ethiopic is the attachment of suffixes to keFi when used as a numeral. These forms keleHU, kele'itd, keleHi

a n d kele'eta, m a s c u l i n e ,

fem., n o m i n a t i v e

and

accusative respectively, are based on the analogy of other numerals with suffixes (Praetorius, §135). Still another case is found in the forms of prepositions with suffixes. The prepositions 'to,' 1? 'unto,' and ^ 'over,' are to be referred to the stems HI?, and (see Stade, §375). Hence 1 with suffixes we have the organic forms T i r . etc. But the language mistook this ' for a formative element, forming the plural, and then added it to the prepositions where it had no place at all. T30S ' after thee,' ' under thee," iV.ri ' between us,' 'among us.' Cf. Barth, Vergleichende Studien, Z D M G 42, 348-358, where also the views of Lagarde, Mittheilungen I 232 are successfully combatted. S o too in Ethiopic all the prepositions receive their suffixes with the insertion of the long vowel £ (see Praetorius, Ethiopic Grammar, §152). This vowel is organic only in Id'la 'upon,' with suffixes (cf. Syr. stem But the language mistook it for a formative element and then applied it analogically in the case of the other prepositions. Another case of this class is found in the form of the feminine plural before suffixes in Hebrew, e. g. WniD-ID. Losing sight of the fact that the ^ was organic only in the masculine U'DID, and

150

AMERICAN

JOURNAL

OF

PHILOLOGY.

that the fem. pi. construct was already expressed by the ending fit, the language adding the ending v- through the influence of the masculine. On this subject compare the remarks of Praetorius, Tigrina Sprache, §99, the substance of which I quote as follows: In several ancient and modern Semitic dialects the pronominal suffixes are found partially or wholly in combination or even union with a nominal plural ending. T h e languages presenting this phenomenon are Hebrew, Aramean, Modern Syriac, Mandean, Amharic, and Tigrina. T h e similar occurrence, however, of this peculiarity in these six languages is purely external and accidental. There are three different grounds and occasions of this confusion of plural ending and pronominal suffix which are still to be distinguished in its occurrence in the different languages. In modern Syriac and Mandean the plural nature of the nominal plural ending was entirely forgotten and abandoned. See Noldeke, Mand. Grammar, §76 and §141. T h e possessive suffix pronouns have incorporated the ending ai, which originally served as the ending of the construct masc. pi. T h e result is that in the case of masculine nouns with pronominal suffixes the singular and plural can no longer be distinguished. T h e same thing is true in modern Syriac. See Noldeke, Modern Syriac Grammar, §37. In Hebrew and Aramean, however, a knowledge of the plural nature of * was retained, in spite of its erroneous application. This is sufficiently shown in the fact that besides its legitimate application in the masculine, it is found only in the plural of the feminine. S o also in Tigrina and Amharic the meaning of the plural ending in combination with a suffix was not forgotten. When, however, we find, in Tigrina, cases where the singular has the suffix -6tk&m or -atom instead of and alongside of the simple -k-drn and -6m, this must be because the plural ending -dt was thought to denote not only the plural nature of the noun to which it was suffixed, and which was its original function, but also the plural nature of the following pronominal suffix. T h e ending -&t, therefore, is capable of a two-fold reference—either to the preceding noun, which is legitimate, or to the following pronominal suffix, where it is not legitimate. A s a result of this confusion k&hen&tMm, for instance, may mean either ' y o u r priest,' or 'your priests.' Sometimes there is a misapprehension of the nature and origin

ANALOGY

IN

THE

SEMITIC

LANGUAGES.

151

of the formative elements of a word. See Dr. Martin Schultze, Zur Formenlehre des semit. Verbs, p. 28, §29. He there points out that the Syriac interjectional n and the Biblical-Aramean prepositional ^ of 3d person sing, imperf. came to be regarded as the pronominal prefix of the 3d person, and as such was prefixed to the plural as well as to the singular. Before Schultze, however (1886), Mr. W . H. Salter Brooks (1883), in the appendix to his treatise "Vestiges of the Broken Plural in Hebrew," pointed out that the Syriac form with n belonged originally only to the singular, and was afterward applied analogically to the plural forms. He says nothing, however, as to the original nature of this n. Another note on the prefix ^ and ! of the 3d person impft, in Aramean and Syriac respectively. As to the Aramean, see the remarks of Kautsch, Grammar, p. 79. The ^ was originally the sign of the optative or precative (compare Assyrian precative particle lit). Afterwards its original signification was lost and it was applied indiscriminately.1 Compare the examples cited by Dr. Rosenberg," Das aramäische Verbum im babylonischen Talmud," Marburg, 1888. Dr. Rosenberg says: is prefixed after the example of the Assyrian, where in the same relations the particle lu is used, to strengthen the assertion. Gradually its force weakened until it became simple assertion and took the place of the prefix \ Because of its relationship with J, however, it changed into the same, and this new form became the exclusive one in Syriac, in Mandean with few exceptions, and in Talmudic was used pretty often." According to Nöldeke, in Lazarus and Steinthal, Zeitschrift für Völkerpsychologie und Sprachwissenschaft, Vol. VII, pp. 403-411 the dual originally denoted a pair (die paarweise Verbindung). This signification is still retained in Hebrew. In Arabic, however, it came to mean simply twoness, and extended itself to all the inflected parts of speech. If this could be maintained it would be an interesting case of the analogical extension of an inflectional element beyond its original bounds. In a review of this article, however, by Dr. Friedrich Müller, " Der Dual in den semitischen Sprachen," Wien, 1875, this position is questioned, and the counter-position maintained, that the Arabic dual form of the verb, qatalä, is the original. Friedrich Müller endeavors to support the position by a comparison with 1

Cf. Haupt, Beitr. z. Assyriologie, I 17.

152

AMERICAN

JOURNAL

OF

PHILOLOGY.

the Assyrian (p. 8). This must be abandoned, however, as the Assyrian form cited is now known to be, not a dual, but the feminine plural. 1 W e need not now examine the details of his argument. On the whole, his reasoning is unsatisfactory, and I do not accept his conclusion (p. 12 and p. 14) that the Arabic dual, in its form and in the extent of its application, represents most faithfully the condition of the parent Semitic in this respect. T h e position of Nöldeke seems to be the correct one. A case of misconception of the nature and significance of an inflectional ending and consequent misapplication is given in Nöld. Neusyrisch. Gramm, p. 218, §102, the use of the ending -ni, which should be used only in the plural, to form the 3d person sing. A similar case is the indiscriminate use of the old Ethiopic Abüna or Abün in modern Amharic as the title of a bishop or a saint (see Praetorius, Amharische Sprache, §76b). For another case of this class see Praetorius, Tigrina Sprache, §131. T h e plural ending -df, which originally was the ending of the feminine plural, has in many cases taken the place of the original masculine ending ~än, especially in the case of adjectives and participles. For the same phenomenon in Tigre compare Schreiber, Manuel de la langue Tigrai, §63. In this connection we must bear in mind, however, that the presumption is that in primitive times the plural endings -im or -in or -Cm, on the one hand, and -&t on the other, were applied more indiscriminately without regard to gender; compare e. g. n^K and D'BÖ. See Barth, Vergleichende Studien, Z D M G 41, 613, 614. In studying the plurals formed by the insertion of h (see Nöldeke, Proc. of the Berlin Acad. 1882, 1178-9; Barth, Vergleichende Studien, Z D M G 41, 621 fg.), we find some instances in which it is inserted probably after the analogy of other older and more original cases. Thus Arabic ummah&t ' mothers' probably after the analogy of an older form abahät' fathers '; cf. the Sabean Tims* and the Syriac (cf. Hebrew ninDS?2 'handmaids') is not therefore a late form. On the contrary it is primitive Semitic. See Barth, Z D M G 41, 624, note 2. See, however, Haupt, SFG 71, and ZDMG 34. See also E. Nestle's paper on El, Elohim, Elöah in Theol. Studien aus Würtemberg, III. Jahrg. (1882), pp. 249 and 255. According to Nestle DTPK is plural to "PK and the singular Hlbx a secondary analogical derivation from 1

5

DTl'pK.

ANALOGY

IN

THE

SEMITIC

LANGUAGES.

153

In Aramaic (Syriac) the h was originally inserted only before the ending -àtha. The few cases inwhich it has been inserted before the ending -in (é) such as NH3Ì? (cf. Sab. TDK, const, pi.), «nan 'father-in-law,' KHD^ 'name,' are probably of a secondary nature and formed after the analogy of forms in -àtha, xnnax, snnort.* The h arose from an original w because of dissimilation. In almost all cases where we find the original stem-consonant w changed into h in the plural, we find that it immediately follows a labial. Hence the labial spirant w was changed to the guttural spirant h. Cases where a labial does not precede are hence to be considered analogical formations. They are comparatively rare, and the different forms appear respectively in only one language ; so Arab, sanakàt, Phoenician nin^H. An interesting and valuable discussion of some analogical formations of this class is given by Barth, ZDMG 42, 341-358. I shall give only the summary of his views, referring to the article itself for the details. The words for heaven, water, life, entrails, face, value, in the several North-Semitic languages have the plural form, while the corresponding forms in South-Semitic are generally singular. Cf. Heb. D W , Phoen. DK>K\ Syr. PPtp, w a f , Psalm. with the Arabic samd', Sabean Ethiopic and Amharic samài. Heb. Syriac TP' K'tD, with Arab, ma', Eth. and Amh. mài. Heb. Syr. |?!> K»n Arab, kaiàt. Hebrew (late, Mishna) or DWO, Arab. mi'an or mìà\ Hebrew D'JB. Late Hebrew DW, Aram. XW. Barth thus explains these plural forms which are found in Hebrew and other North-Semitic languages. They are all derived from stems I"1? and In the case of the words for heaven and water the original forms were probably samài and mài. These were shortened to samài and mài, thus exactly resembling the usual construct, pi. masc. Also when these forms received suffixes they presented the same appearance as the plural with suffix. Hence the language mistook the form for a plural, and formed the corresponding plural absolute O'W and 1 T h e supposition of Stade, §182, that the insertion of the H is to give biconsonantal stems the appearance of tri-consonantalism cannot be maintained. See Barth, Z D M G 4-', 625.

154

AMERICAN

JOURNAL

OF

PHILOLOGY.

In the same manner, from the forms with suffixes, Tin. which, although really singulars, looked like plurals, were formed the absolute plurals D'OS, D W , S o perhaps also Bibl. Aram. TIG 'breast,' and Hebrew ' blood-guiltiness.' According to Barth (Nominalbildung, §123), the prefix ma of the Arabic passive participle of the first form is due to the influence of the analogy of the passive participles of the other forms. This is a case of the analogical extension and application of a formative element beyond its original bounds. Sometimes analogical influence of this kind produces great mixtures of forms. T h u s the G r e e k suffix -dpios is attached to genuine Syriac words (Noldeke, §140, p. 77). Here also may be classed such formations as are mentioned in Noldeke's Mandean Grammar, p. 86 fg. achuun, achni " we " forms. achtun " y o u , " b y inserting the ch, a contamination of form. Then, as from achtun there was formed achtdchun by the possessive suffix of the 2d pi., this again produced the formation achnochun for the first person. HISTORICAL

AND

BIBLIOGRAPHICAL

NOTE.

N o systematic comprehensive treatment of analogy in the Semitic languages has yet appeared. In nearly all the recent treatises on the grammar, phonetics, and morphology of these languages frequent use is made of the principle of analogy to explain the different phonetic and morphological phenomena of these languages. So, for example, Bickell (Grundriss der hebraischen Grammatik, Leipzig, 1869), English translation, Outlines of Hebrew Grammar (Leipzig, 1877), on page 42, §47, notes the fact that " t h e remarkable transition of i, which has arisen from a, to i in the hiphil has perhaps originated according to an erroneous analogy from the conjugation of the verbs mediae v, y, where this i is phonetically legitimate." A n d on page 49 of the same work he suggests that many Semitic stems which at present have three stem-consonants originally had only two such consonants, and were made tri-consonantal b y the insertion of an auxiliary sound through a pervasive analogy. A l s o in the following among many other passages of the same work he points out cases of the occurrence of analogy. O n page

ANALOGY

IN

THE

SEMITIC

LANGUAGES.

155

66, §86, he notes that after a false analogy the ending ä, to indicate direction, is added to a plural noun in the case of HD^B'S ' unto Chaldea,' Ezek. 16. 29. On page 70, §93, it is observed that before suffixes the plural endings of the masculine are superadded after the feminine plural ending according to a false analogy peculiar to Hebrew; cf. TP' 10 ' thy horses,' and TniDlD ' thy mares.' On page 100, §136, it is observed that the feminine singular perfect of verbs tertiae infirmae (galayat, gal&t) is usually expanded to after the analogy of strong verbs. In " D i e sumerischen Familiengesetze von Dr. Paul Haupt," Leipzig, 1879, we have many discriminating and valuable observations on the occurrence of analogy in Assyrian and also in the other Semitic languages. See page 52, note 10, where it is pointed out that in Assyrian izzazü (3d plur. pres. kal stem naz&zu 'to settle, to stand') instead of the organic form in&zazu is based on the analogy of illakü 'they g o ' (stem alaku), the analogy being induced by the form of the imperfects izzizü and illikü. (On these forms cf., however, Del. Ass. Grammar, §100.) In the same place it is also pointed out that Hebrew (grdform anaku) and 'i^Pi? are based on the analogy of ^ the vocalization of .the final syllable for anaku and qataltu, while qataltu is for qatalku under the influence of the 2d person qatalta and qatalti. Unfortunately the treatise which is there promised us on "Associative Neubildungen im Bereich des Semitischen" has not yet made its appearance. On page 66, note 3, of this work it is pointed out that many forms of verbs 1"B (Assyrian) such as uUbila ' I brought,' iiiesib ' I placed,' use§ü 'they led out,' are based on the analogy of forms from stems On page 73 middle analogy is pointed out in the forms muttäti, plural of muttu for mundii, similar to the analogy observed in the Hebrew forms nin1?"^ plural of n V i ' door,' and ¡"iiriB'i?, plural of n f i j 'bow.' T w o articles by the same author in the first volume of Hebraica (Assyrian Phonology and Wateh-ben-Hazael), as also several articles in the first number of Beiträge zur Assyriologie, Delitzsch und Haupt, Leipzig, 1889, contain much valuable material of the same kind which has been utilized for this present treatise.

156

AMERICAN

JOURNAL

OF

PHILOLOGY.

In Stade's Hebrew Grammar, Leipzig, 1879, many individual cases of analogy are pointed out in the discussion of the various forms. T h e greater number of these cases are cited and arranged under their respective classes and divisions in the subsequent part of this treatise. On pages 110-114 of his grammar Stade treats of the various ways in which original bi-consonantal stems have become triconsonantal under the influence of the analogy of those stems which originally had three stem-consonants. For the rest the principal individual cases of analogy which he points out are those cases where a weak or irregular verb-stem is treated after the analogy of a strong stem, or where the different classes of weak verb-stems are confused, for instance, confusion of stems and X"1?, §i43e, note 1. It is hardly necessary to say that in the excellent grammars of Theodor Nöldeke (Syrische Grammatik, Leipzig, 1880; Neusyrische Grammatik, Leipzig, 1868; Mandäische Grammatik, Halle, 1875) we have a comparatively full presentation of nearly all the analogical phenomena of the respective languages of which he treats. These different formations, as given by Nöldeke, will be found cited or referred to in the subsequent pages of this thesis. The same remarks to some extent apply also to the grammars of Kautzsch (Grammatik des Biblisch-aramäischen, Leipzig, 1884), Praetorius (Grammatik der Tigrina-Sprache, Halle, 1871; Die Amharische Sprache, 1879; ä t h i o p i s c h e Grammatik, Leipzig, 1886), and Delitzsch (Assyrische Grammatik, Berlin, 1889). Considerable valuable material is also contained in various essays on different topics of comparative Semitic grammar recently published. See especially Nöldeke in Z D M G 37, 525, Die Verba 'W im Hebräischen; Vol. 38, 407, Die Endungen des Perfects; Barth, Die Nominalbildung in den semitischen Sprachen, Leipzig, 1889, also his Vergleichende Studien in Z D M G 41,603; 42, 413 ; 43, 177. T h e material of these articles has been cited and discussed in this paper. There is not much that is new in Wright's Comparative Grammar of the Semitic Languages, recently published (New Y o r k , 1890). ABEL H. HUIZINGA.